alii1

Embed Size (px)

Citation preview

  • Please citeresources,

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages12Accident Analysis and Prevention xxx (2011) xxxxxx

    Contents lists available at ScienceDirect

    Accident Analysis and Prevention

    journa l homepage: www.e lsev ier .co

    Psycho ofharassm h a

    Rebecca Chra Work & Stres ersityb Johannes Gut

    a r t i c l

    Article history:Received 6 DeReceived in reAccepted 5 Ap

    Keywords:Psychosocial sWork psychosocial riskWork stressOccupational SafetyBullying and harassment

    neder psyhe Joorkas deempl

    organizational PSC was negatively associated with workplace bullying and harassment (demands) andin turn psychological health problems (health impairment path). PSC was also positively associated withwork rewards (resources) and in turn work engagement (motivational path). Accordingly, we found thatPSC triggered both the health impairment and motivational pathways, thus justifying extending the JD-Rmodel in a multilevel way. Further we found that PSC, as an organization-based resource, moderated thepositive relationship between bullying/harassment and psychological health problems, and the negativerelationship between bullying/harassment and engagement. The ndings provide evidence for a multi-

    1. Introdu

    In manyassessable ddirect and iCompensatnational pristress resulhazards asnizational msocial harmstress, suggand CooperBelgium,Defor employet al., 2008)cation and mcauseof the

    CorresponE-mail add

    0001-4575/$ doi:10.1016/j.this article in press as: Law, R., et al., Psychosocial safety climate as a lead indicator of workplace bullying and harassment, jobpsychological health and employee engagement. Accid. Anal. Prev. (2011), doi:10.1016/j.aap.2011.04.010

    level model of PSC as a lead indicator of workplace psychosocial hazards (high demands, low resources),psychological health and employee engagement, and as a potential moderator of psychosocial hazardeffects. PSC is therefore an efcient target for primary and secondary intervention.

    2011 Elsevier Ltd. All rights reserved.

    ction

    countries work stress is considered a preventable risk-isease (Clarke and Cooper, 2000). Given the signicantndirect costs associated with it (Australian Safety andion Council; ASCC, 2006), job stress is increasingly aority occupational health and safety (OHS) issue. Workts from prolonged exposure to workplace psychosocialpects of the work environment, work design, and orga-anagement which potentially cause psychological and(Cox et al., 2000). The slow accident effect of workests that latent causes are readily identiable (Clarke, 2000). OHS legislation in countries such as Australia,nmark, theNetherlands, and Sweden species the needers to take action against psychosocial hazards (Ertel. In accord with the hierarchy of controls, the identi-anagement of more distal hazards and risks, so calledcauses,will provideamore reliable, efcient andeffec-

    ding author. Tel.: +61 8 83022775; fax: +61 8 83022956.ress: [email protected] (M.F. Dollard).

    tive control strategy (Dollard, 2011). In this paper we propose thatpsychosocial safety climate (PSC) is a distal upstream determinantof both hazards and job stress related outcomes.

    Psychosocial safety climate is dened as organizational poli-cies, practices, and procedures for the protection of workerpsychological health and safety (Dollard and Bakker, 2010, p. 580).We propose that PSC is a lead indicator of commonly identiedpsychosocial hazards at work. In particular we focus on the work-related hazards (or demands) harassment and bullying as theycomprise on average 24% of all accepted mental stress claims inAustralia (20072008) (Productivity Commission, 2010). Our the-oretical analysis brings together concepts from the safety science,work stress, and organizational psychology literatures, and posesPSC as an upstream antecedent to workplace psychosocial hazardsin the form of (a) social and emotional demands (i.e., bullying andharassment) and (b) low job resources (i.e., supervisor support,job rewards, and procedural justice). We also explore the amelio-rative effects of PSC as a moderator of the deleterious effects ofjob demands. The antecedent and ameliorative roles of PSC in thismodel, approximate the primary prevention, and secondary pre-vention roles of PSC respectively. Moreover, it is important thatorganizational performance andproduction goals are not neglected

    see front matter 2011 Elsevier Ltd. All rights reserved.aap.2011.04.010social safety climate as a lead indicatorent, job resources, psychological healt

    Lawa, Maureen F. Dollarda,, Michelle R. Tuckeya,s Research Group, Centre for Applied Psychological Research, School of Psychology, Univenberg-University, Mainz, Germany

    e i n f o

    cember 2010vised form 16 March 2011ril 2011

    afety climate

    a b s t r a c t

    Psychosocial safety climate (PSC) is deprocedures for the protection of workment practices. PSC theory extends torganizational level PSC determines wand work engagement. Our sample wcomprised 30 organizations, and 220m/locate /aap

    workplace bullying andnd employee engagement

    istian Dormannb

    of South Australia, Adelaide, South Australia, Australia

    as shared perceptions of organizational policies, practices andchological health and safety, that stem largely from manage-b DemandsResources (JD-R) framework and proposes thatconditions and subsequently, psychological health problemsrived from the Australian Workplace Barometer project andoyees. As expected, hierarchical linear modeling showed that

  • Please cite a lead indicator of workplace bullying and harassment, jobresources, 11), doi:10.1016/j.aap.2011.04.010

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages122 R. Law et al. / Accident Analysis and Prevention xxx (2011) xxxxxx

    in the pursuit of health improvements. Therefore, we explore therole of PSC in stimulating psychological health in combinationwithwork motivation outcomes.

    1.1. Psycho

    The conupon persporganizatiospecic comfrom psychIt reects mhealth andhealth as o2010). Likeerty of theof individucommitmen(Dollard anthe idea thronments co-worker2008). As suals include hreporting ping all layerand Bakker

    The theothat of safettors and psyresearch haphysical injuand industrGrifn, 200climate to s2000; Zohaand industrNeal and Grbeen the idcal health ahazards and

    Consisteclimate (i.e.injury. Applof physicalmay be conconditions (sure, and chabout our stharassmentlack of supedural justicunderstoodand Luria, 2theory for P

    1.2. PsychoDemandsR

    A basiclargely fromical health,(proximal)ers. To buiDemandsR2001), whic

    sychosofety clim

    2nizao

    Health erosion path

    k anouti,impae jobrgy ral, soed pfeli are per eof joe themssu007;seco

    al poorkmerng joand

    s a rey pramework (see also Dollard and Bakker, 2010; Idris et al., in

    ealth erosion pathway: workplace bullying and harassment

    ur current PSC study model, job demands are operational-y workplace bullying and harassment. Here, bullying isd by three core features: duration (the offensive behaviorsregularly and repeatedly over a period of six months), vic-tion (the victim experiences difculties defending him or), and a power differential (the behaviors occur betweenrties of disproportionate power) (see Lindstrm et al., 2000).ment is identied as perceived sexual harassment, discrim-treatment, and psychological humiliation (Richman et al.,Given the theoretical interrelations of these forms of vic-tion (Aquino and Lamertz, 2004; Bowling and Beehr, 2006),vision similar antecedents and consequences. Exposure toorms ofworkplace victimization carries high risk of harm foruals, for example depression, anxiety, irritability (MayhewcCarthy, 2005), poor cardiovascular health (Tuckey et al.,this article in press as: Law, R., et al., Psychosocial safety climate aspsychological health and employee engagement. Accid. Anal. Prev. (20

    social safety climate

    ceptual theory of psychosocial safety climate drawsectives from the work stress, psychosocial risk, andnal climate literatures (Dollard, 2011). PSC is a facet-ponent of organizational climate relating to freedomological harm at work (Dollard and Bakker, 2010).anagement commitment to workers psychological

    the priority they give to safeguarding psychologicalpposed to production demands (Dollard and Bakker,organizational climate, PSC is conceived as a prop-

    organization, consisting of the aggregated perceptionsals within that organization regarding managementt to protecting their psychological health and safety

    d Bakker, 2010). The PSC construct stems largely fromat individuals ascribe meaning to their work envi-their working conditions, management systems, pay,

    relationships, and treatment equity (see James et al.,ch, ways in which PSC can become visible to individu-avingwell developed communication systems (e.g., foroor psychological health at work) and actively involv-s of the organization in work stress prevention (Dollard, 2010).retical basis of psychosocial safety climate is similar toy climate but focusesmore sharply on psychosocial fac-chological health. In its 30-year history, safety climates focused on accidents, errors, and disasters resulting inry, and the high direct and indirect costs to personnely in conditions where safety climate is poor (Neal and6). Substantial evidencehasbeenamassed linking safetyafetybehavior andperformance (Clarke, 2006; Flinet al.,r, 2010), physical injury (Silva et al., 2004; Zohar, 2010),ial accidents and errors (Hofmann and Stetzer, 1996;ifn, 2006). Missing in the literature until recently hasentication of a specic safety climate for psychologi-nd safety that may be a lead indicator of psychosocialpsychological health at work.

    nt with the safety climate literature, we propose that, PSC) precedes the conditions that lead to psychologicalying safety science theory in a novelway, the likelihoodinjury and in addition psychological injury at work ceived as the joint outcome of proximal factors, unsafesocial/technicalhazards), unsafe acts, cumulativeexpo-ance variations (Reason, 1997; Zohar, 2010). Thinkingudy in safety literature terms, exposure to bullying andat work represents a hazard or unsafe condition, as arervisor support, inadequate job rewards, and low proce-e. In the work stress literature, unsafe hazards may beas high job demands and low job resources (c.f. Zohar005). Merging these perspectives forms the crux of theSC.

    social safety climate extending the Jobesources model

    premise in our PSC theory is that PSC, stemmingmanagement practices regarding worker psycholog-acts as a precursor to psychosocial hazards the

    job demands and job resources experienced by work-ld a model for our study we rst turn to the Jobesources (JD-R) model of work stress (Demerouti et al.,h describes how job demands and resources are linked

    Psa

    LevelOrga

    to worDemerhealthmanagof enephysicsustain(Schaument a(Semmvarietyactivatprobleet al., 2

    Thevationhigh wand Dereducigrowthtions arisks bJD-R frpress).

    1.3. H

    In oized bdeneoccurtimizaherselftwo paHarassinatory1996).timizawe enthese findividand MH6

    H5

    H3, H4

    H4

    H2

    H1, H2

    cial ate

    Bullying/Harassment

    Psychological health

    problems

    ResourcesRewardsJusce

    Supervisor support

    Engagement

    nLevel 1Individual

    Movaonal path

    Fig. 1. Study model.

    d health outcomes through two processes (Bakker and2007; Hakanen et al., 2008b; see Fig. 1). The rst is theirment process which explains how sustained effort todemands contributes to strain through the exhaustion

    eserves (Schaufeli and Bakker, 2004). Job demands arecial, or organizational aspects of the job that requirehysical, psychological, cognitive, or emotional effortndBakker, 2004). Demands such as bullying andharass-articularly stressful as they directly threaten the selft al., 2005). Studies have consistently shown that ab demands (e.g., emotional demands, work pressures)health impairment process, leading to increased healthchaspsychologicaldistress (Bakkeret al., 2004;DollardHakanen et al., 2006).nd motivational process (see Fig. 1) describes the moti-tential of job resources to stimulate outcomes such asengagement and increased work performance (Bakkerouti, 2007) extrinsically (by aiding goal attainment andb demands), and intrinsically (by stimulating personallearning) (Bakker et al., 2003). We argue that PSC func-ference point for the presence and level of psychosocialeceding these two processes, thereby augmenting the

  • Please cite a leresources, 11), d

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages12R. Law et al. / Accident Analysis and Prevention xxx (2011) xxxxxx 3

    2010) and traumatic stress (Bond et al., 2010). Moreover, organi-zations suffer as a result of workplace victimization in terms ofstaff turnover, sickness absence, workers compensation, reducedmorale and motivation, and diminished productivity (Hoel et al.,2003). Giveantecedent

    In organworker psythat low levrelationshiphypothesisate and sustbehavior suand complaPSC is low)(Ashford, 1bullying maing order oTuckey et ament assumand Harveypractices anbe deterredaccordinglylying and hBeginning wprotect empand proceduprocess, ltand superviLuria, 2005et al. (2010police ofclower bullyhealth erosthe study m

    Hypothesisworkplace b

    Hypothesispsychologicwith job dedemandswlems in a m

    1.4. Motivarewards and

    As mentductivity anemployeeschosocial rmodel, thezational rewto perceivetion, perforand employ(Cohen-Chaemployeesinstrumenttomonetary1996).

    The linkbe explaineory, workerinvested in

    are more likely to reciprocate through motivation and engagementat work (Blau, 1964; Maslach et al., 2001; Schaufeli and Bakker,2004). Employees in effect repay the organization in response toresources received (Saks, 2006). Indeed, procedural justicehasbeen

    to dezatiop behperto evaraositiighadethatitiveSC or supn vasfactions.cinguencicatoted ill-beivatikkeryee eces. W

    hesisourcisor

    hesismenrwod wo

    ditio

    alsoabo

    empohr end hoptiohe avdistrrce tnd hue tsup

    ay nSimitweenal sdividderaaumxt hy

    hesisetwems. Un bureduthis article in press as: Law, R., et al., Psychosocial safety climate aspsychological health and employee engagement. Accid. Anal. Prev. (20

    n these signicant costs it is vital to understand thes of bullying and harassment at work.izations characterized by high levels of PSC, wherechological health and safety is protected, we expectels of bullying and harassment will be observed. Thiscan be understood through the work environment

    (Salin, 2003), whereby poorworking environments cre-ain conditions that are conducive to bullying. If deviantch as bullying and harassment goes unacknowledgedints are not acted upon (i.e., as could be expected when, such behaviors effectively become institutionalized994; Liefooghe and Davey, 2001). In low PSC contexts,y be perpetrated in a top-down fashion (e.g., in a peck-f seniority in hierarchical cultures; Paice et al., 2004;l., 2009). By contrast, in high PSC contexts, top manage-e responsibility for harassment and bullying (Heames, 2006), through the enactment of relevant policies,d procedures. Therefore bullying and harassment mayif top management adopts a zero-tolerance stance and,, the impetus for safeguarding employees from bul-arassment ows primarily from senior management.ith senior management priority and commitment toloyees frompsychosocial harm, organizational policiesres (in this case, PSC) operate in a multilevel top-down

    ering down to lower levels where middle managementsors maintain and transmit these standards (Zohar and). In support of the directionality of our theory, Bond) found that low PSC predicted workplace bullying iners over time, whereby high PSC was associated withing over time. Therefore, in line with the extendedion process of the JDR model we propose (see Fig. 1,odel):

    1. Organizational PSC will be negatively related toullying and harassment.

    2. Organizational PSC will be negatively related toal health problems through its negative relationshipmands (bullying and harassment). In other words, jobill carry theeffect of PSContopsychologicalhealthprob-ediated process.

    tional pathway: procedural fairness, organizationalsupervisor support

    ioned, we believe that PSC should also relate to pro-d performance goals. To achieve organizational goals,should be provided with appropriate physical and psy-esources (Hobfoll, 1986). Within our extended JD-Rresources of interest are procedural fairness, organi-ards, and supervisor support. Procedural justice refers

    d fairness of procedural decisions, such as pay selec-mance evaluation, promotion (Lind and Tyler, 1988)ee perceptions of how organizations allocate resourcesrash and Spector, 2001). Supervisor support refers toperceived social support from supervisors, including

    al andemotional support.Organizational rewards relate, esteem, and status control (e.g., job security) (Siegrist,

    between job resources and motivational outcomes cand by social exchange theory. According to this the-s who perceive or valuate their organization as beingtheir well-being, through adequate resource allocation,

    linkedorganizenshiand jobrelatedViswesbeen p

    In hrequirenizantof poshigh Pgreatenizatioof saticonditresourconseqan indassociative wePSC actandBaemploresour

    Hypotjob ressuperv

    HypotengageinothePSC an

    1.5. Ad

    Wemationwhichcues (Llying aactionoffset tlogicala resoulying aharm dwill beand mdo so.linkbeemotiothan inPSC mopost-trthe ne

    Hypotship bproblebetweewill bead indicator of workplace bullying and harassment, joboi:10.1016/j.aap.2011.04.010

    clines in job satisfaction (Folger and Konovsky, 1989),nal commitment (Daly and Geyer, 1995), helpful citi-aviors (Moorman, 1991; Organ and Moorman, 1993),

    formance (Gilliland, 1994); supervisor support has beennhanced performance (Rhoades and Eisenberger, 2002;n et al., 1999); and rewards and supervisor support havevely related to engagement (Demerouti et al., 2001).PSC contexts, managers will be aware that workersquate resources to complete job tasks and will be cog-not having enough resources leads to reduced levelswork emotions (Schaufeli and Bakker, 2004). Withinrganizations we can expect higher procedural justice,ervisor support and greater rewards because the orga-lues the positive well-being of its workforce (i.e., levelsion, engagement) and thus creates optimum workingSimilarly, managers will understand that inadequatemay lead to negative reactions and counterproductivees (Spector et al., 2006). In other words, PSC should ber of adequate job resourcing within organizations, andndividual level motivational processes that foster posi-ing outcomes. Previous research has found support forng themotivational pathway of the JD-Rmodel; Dollard(2010) and Idris et al. (in press) found that PSCpredictedngagement through its relationship to a change in jobe therefore propose:

    3. Organizational PSC will be positively related toes (i.e., procedural justice, organizational rewards, andsupport).

    4. Organizational PSC has a positive effect on workt through its positive relationship with job resources;rds, job resourceswillmediate the relationshipbetweenrk engagement.

    nal pathways: moderation effects of PSC

    expect a safety signal effectwherebyPSCprovides infor-ut possible resource options in the environment uponloyees can act to provide respite or relief from dangert al., 2007). When danger cues such as workplace bul-arassment are present, the PSC safety signal indicatesns (e.g., utilization of available emotional resources) toersive stimuli, and to avoid thedevelopment of psycho-ess (see Lohr et al., 2007). In otherwords PSC representshat can be used deal with demands. In relation to bul-arassment, for example, when workers feel safe fromo a strong psychosocial safety climate, they know theyported to cope with any negative treatment they faceot need to draw so heavily on their own resources tolarly, Bacharach and Bamberger (2007) found that thenre-ghter critical incident involvement andnegativetates varied as a function of station/unit-level (ratherual) job resources. Previous research showedunit-levelted the relationship between bullying/harassment andatic stress symptoms (Bond et al., 2010). This leads topothesis:

    5. Organizational PSC will moderate the relation-en bullying and harassment and psychological healthnder conditions of high PSC, the positive relationshipllying/harassment and psychological health problemsced. Conversely, in low PSC climates, reporting bullying

  • Please cite a leresources, 11), d

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages124 R. Law et al. / Accident Analysis and Prevention xxx (2011) xxxxxx

    or harassment may lead to an exacerbation of the problem, due tovictim blaming and scapegoating.

    Additionally, there is an abundance of evidence on the nega-tive affectivharassmenttional efforreserves, bself-associa(Semmer etfer. We proimpact on e(Dollard anresources, cthe social anment particWhen suppdemands, was meaning(absorption(vigor). In a

    Hypothesistionship beunder condbullying/ha

    1.6. Level o

    There iswhether thnizational oshared perctures (Nealat an indivi(James et aship betweonly 20% ofilarly Clarkpsychologiccomes reveThis issue irecent metabe conductgroup level

    Psychosof the orgabecause it iswith safetySenior mandecisions scedures, anproductivitfor the PSCBakker (20samples, wperceptionsto provide fmodel in thlevel pheno

    2. Method

    2.1. Particip

    The samdomly selec

    were collected from individuals in all sectors (private, government,non-government organizations) and professions within the work-force. Prospective participants were individuals in the householdwith a valid telephone connection, who most recently had their

    aywithin the household, between the ages of 18 and65 yearspaid employment, not self-employed, and agreed to provideation for the survey. Those who did not meet all of the abovewere excluded from the study. A computer assisted tele-

    interviewing (CATI) techniquewas used. This techniquewasin order to maximize response rates, minimize data collec-e anderrors, enable data collection fromEnglish as a second

    ge participants, and to ensure a representative sample. Anuctory letter was sent to potential participants homes andof those who responded to the telephone survey indicatedey hhe Uitteeotal,Word etres

    Thecs (207.

    ymen.1

    ticipad for. Werome da43e n(ma

    6, p=yearpulatnentart-t(5.9

    didThegestmenunitying s19)3 (9respexpezatio

    esign

    rese, bas

    wereabilit

    in paiployf thatrget pthis article in press as: Law, R., et al., Psychosocial safety climate aspsychological health and employee engagement. Accid. Anal. Prev. (20

    e-cognitive consequences of exposure to bullying and, as noted above. As a result of the social and emo-t required and the corresponding depletion of energyut more importantly, due to the direct threat to theted with being victimized by bullying and harassmental., 2005), employee work engagement is likely to suf-pose, however, that PSC should mitigate this negativengagement. PSC acts as an organization-based resourced Bakker, 2010) that, over and above personal and joban be harnessed to help affected workers cope withd emotional demands of bullying/harassment. Engage-ularly reects how employees experience their work.orted by a strong PSC to manage bullying/harassmentorkers may thus still be able to experience their workful (dedication component of engagement), interesting), and something to which they wish to devote effortccordance we propose:

    6. OrganizationalPSCwillmoderate thenegative rela-tween bullying/harassment and engagement. That is,itions of high PSC the negative relationship betweenrassment and engagement will be reduced.

    f operationalization

    considerable confusion in the literature regardinge safety climate construct is a property of the orga-r the individual. Despite the term climate implyingeptions that may reect organizational or group fea-and Grifn, 2006), most studies operationalize climatedual level, and refer to this as the psychological climatel., 2008). A recent review of 35 studies of the relation-en safety climate and safety performance showed thatstudies used group level analysis (Clarke, 2006). Sim-

    es (2010) meta-analysis of the relationship betweenal climate, safety climate, and individual safety out-aled only 7% of the studies used group level analysis.s not specic to safety climate, as demonstrated in a-analysis of organizational climate research that had toed at the individual level because of a low number ofstudies (Parker et al., 2003).ocial safety climate is mainly conceived as an attributenization. It is theorized to vary across organizationslargely inuencedby seniormanagement, as is the caseclimate (cf. Huang et al., 2007; Zohar and Luria, 2005).agement are responsible for creating PSC via executiveuch as budgets, resource allocation, policies and pro-d corporate priorities (e.g., the competing demands ofy and prot versus stress prevention). To date, evidencemodel is derived from the seminal work by Dollard and10) and Bond et al. (2010), using single occupationalhere PSC is operationalized by aggregating individualabout workgroup PSC to the workgroup level. In orderurther support for the utility and pervasiveness of theis study we operationalized PSC as an organizationalmenon, and studied it across organizations.

    ants and procedure

    ple consisted of Australian income earners from ran-ted households from the state of South Australia. Data

    birthdold, ofinformcriteriaphonechosention timlanguaintrod63.6%that thfrom tComm

    In ttralian(Dollarsample38.4%.statistiJuly 20emplogures

    Parworkedetailspants finto th(rangeple. Thgender(1) = .32554the popermanent pand 13resultp= .07.the largoverncommRegardtion (5and 20(1.4%)Not unorgani

    2.2. D

    Thedesign

    1 Dataand probworkedthose eminverse oin the taad indicator of workplace bullying and harassment, joboi:10.1016/j.aap.2011.04.010

    ad received the letter. The study had ethics approvalniversity of South Australia Human Research Ethics.there were 1134 participants who completed the Aus-kplace Barometer Questionnaire (AWBQ2009) via CATIal., 2009; Dollard and Skinner, 2007). The overall

    ponse rate was 31.2% and the participation rate wasdata were weighted to the Australian Labour Force006) released by Australian Bureau of Statistics in

    The number of people not self-employed and in paidt by age, group and sex was determined from these

    nts were asked to name the organization that they; N=1043, 91.9% of participants provided employermatched the data by employer, and selected partici-

    organizations forwhichwehadat least fourparticipantstaset. This resulted in a sample of 30 organizations3 respondents, M=7.17) with N=220 in the nal sam-al sample was representative of the population byles, N=99, 45%; females, N=117, 53.2%), Chi-square.36. Most participants (N=146, 66.3%) were betweens of age; this age distribution was representative ofion, Chi-square (6) =2.86, p= .83. The majority workedfull-time (N=113, 51.4%); 48 (21.8%) worked perma-ime; 38 (17.6%) worked on a causal/temporary basis%) were on a xed contract (other N=3, 1.4%). This

    not differ from the population, Chi-square (4) =8.79,re was a wide spread of industries in the sample, withindustries represented being education (N=59, 27%);t administration and defence (N=42, 19%); health andservices (N=38, 18%); and retail trade (N=29, 13.2%).ize of organizations, 4 (2%) worked in a small organiza-; 10 (4.5%) worked in a medium organization (20200);2.3%) worked in a large organization (200+), with 3onding dont know; Chi-square (3) =54.12, p< .001.ctedly, our sample is more likely to represent largerns.

    arch design was a cross-sectional, non-experimentaled on self-report data.

    collected between June and August 2010 and weighted by age, sexy of selection for those aged 18 years and over in the household whod employment, but were not self-employed, to reect the structure ofed full or part time for an employer in SA. The sample weight is thepersons selection probability, and signies the number of individualsopulation that the sampled individual represents.

  • Please cite a leresources, 11), d

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages12R. Law et al. / Accident Analysis and Prevention xxx (2011) xxxxxx 5

    2.3. Measures

    The measures used were drawn from the scales of the Aus-tralianWorkplace Barometer (AWBQ2009) developed at the Centreof Applied(Dollard et

    Controlsandgender,variables mand outcom1990).

    Psychososcale (Dollacontained fmitment (e.of an emplagement prpsychologicorganizationhere aboutand organizbecome invwere measudisagree) tocomposite sthe 12-itemitem scaleconstruct v

    2.3.1. DemaOrganiza

    seven itemsharassmentnic, being sbeing physia 5-point Lioften/alway

    Workpladrawn fromSocial Factolying is a prlabel sometrepeatedlyexperiencebullying if tict of thewere askedplace durin(yes=1, no=

    We creadardizing bmeasure wconstant anskewness (sof variance

    2.3.2. DemaOur mai

    demand coing/harassmthe JCQ (Kave items edemands wlift heavy loworked in tusing four i

    2.0; www.jcqcenter.org, 2009) with a sample question, Does yourwork put you in emotionally disturbing situations? All items weremeasured on a four point Likert scale, ranging from 1 (stronglydisagree) to 4 (strongly agree).

    Resoucedur. Thandny/os tored oto 4ial suon tisor-poinly aganizahe Efll me I deg fro

    Psychcholosslerstionf anxfour-u feeangi10 ttionh Bu

    is:red os); agemShoreewornthumerscomangi

    ta a

    ordeoff aall sn fa

    st acted ftor andenpearen thar mnbusvaricrossith

    ssistthis article in press as: Law, R., et al., Psychosocial safety climate aspsychological health and employee engagement. Accid. Anal. Prev. (20

    Psychological Research, University of South Australiaal., 2009).were age, gender, and income. We controlled for ageand incomeas aproxy to socioeconomic status, as theseay confound relationships between job characteristicse variables (de Jonge et al., 1999; Karasek and Theorell,

    cial safety climate was measured using a 12-itemrd and Kang, 2007; Hall et al., 2010). The measure

    our subscales, each with three items: management com-g., Senior management acts decisively when a concernoyees psychological status is raised), = .91; man-iority (e.g., Senior management considers employeeal health to be as important as productivity), = .90;al communication (e.g., There is good communicationpsychological safety issues which affect me), = .76;ational participation (e.g., Employees are encouraged toolved inpsychological safetymatters),= .80. All itemsred on a 5-point Likert scale, ranging from 1 (strongly5 (strongly agree) and were added together to form acale. High scores reect high levels of PSC (= .94). Bothscale (= .89; Hall et al., 2010) and the abbreviated 4-

    (= .91; Dollard and Bakker, 2010) have shown goodalidity when assessed at the team level.

    ndstional harassment in theworkplacewasmeasuredusing, based on the Richman et al. (1996) scale, canvassingin the following areas: sexual, gender, racial or eth-worn at, being humiliated in front of others, and/orcally assaulted/threatened. Items were measured usingkert scale, ranging from 1 (very rarely/never) to 5 (verys), = .68.ce bullyingwas measured using the following denitiontheGeneralNordicQuestionnaire forPsychological andrs at Work (QPSNordic) (Lindstrm et al., 2000): Bul-oblem at some work-places and for some workers. Tohing as bullying, the offensive behavior has to occurover a period of time, and the person confronted has todifculties defending himor herself. The behavior is notwo parties of approximate equal strength are in con-conict of incident is an isolated event. Respondents: Have you been subjected to bullying at the work-g the last 6 months? answered with a yesno response2).

    ted a composite bullying/harassment measure by stan-oth scores and adding them together. The resultantas skewed (skewness =1.79, S.E. = .17), so we added ad log transformed the measure, which reduced thekewness =1.04, S.E. = .16). This improved homogeneityin the Level 1 measure.

    nd covariatesn focus was on bullying/harassment but we includedvariates to underscore the importance of bully-ent over and above these measures. Scales were fromrasek et al., 1998): (1) work pressure was assessed by.g., My job requires working very hard; (2) physicalere assessed by ve items e.g., I am often required toads on my job; (3) work hours were assessed as hourshe past 7 days; and (4) emotional demands was assessedtems from the new Job Content Questionnaire 2.0 (JCQ

    2.3.3.Pro

    JCQ 2.0designcompatunitiemeasuagree)

    Socbasedsupervon a 4(strong

    Orgfrom tering aprestigrangin

    2.3.4.Psy

    the Keof quelevel orecentdid yoscale, rWe log

    EmoMaslacple itemmeasu(alway

    EngScale were thAt myI am eam imform ascale, r

    2.4. Da

    InPodsakwhereysis. Tethe raccouneral facindepenot ap

    Givcal line(RaudeLevel 1ables aassist w(b) to aad indicator of workplace bullying and harassment, joboi:10.1016/j.aap.2011.04.010

    rcesal justice was measured by four items based on theis measure tapped into the perceived fairness of theimplementation of procedural decisions (e.g., In myrganization, procedures are designed to provide oppor-appeal or challenge a decision.) The items weren a 4-point Likert scale, ranging from 1 (strongly dis-

    (strongly agree); = .82.pport supervisor support was measured by three itemshe JCQ 2.0 which comprised statements such as Myis helpful in getting the job done. Items were measuredt Likert scale, ranging from 1 (strongly disagree) to 4ree); = .83.tional rewards were measured by four items derivedfort Reward Imbalance (ERI; Siegrist, 1996): Consid-y efforts and achievements, I receive the respect andserve atwork. Itemsweremeasuredona4-point scale,m 1 (strongly disagree) to 4 (strongly agree); = .65.

    ological health problems and work outcomesgical distress was measured using the full 10 items from10 (K10; Kessler and Mroczek, 1994). The scale consistss on non-specic psychological distress, including theiety and depressive symptoms experienced in the mostweek period (e.g., In the past 4weeks, about howoftenl nervous?). Items were measured on a 5-point Likertng from 1 (none of the time) to 5 (all of the time); = .84.ransformed this scale due to skewness.al exhaustion was measured using 5 items from thernout Inventory (MBI; Schaufeli et al., 1996). An exam-I feel emotionally drained from my work. Items weren a 7-point Likert scale, ranging from 1 (never) to 7= .87.entwasmeasured using the UtrechtWork Engagementrtened Version (UWES-9; Schaufeli et al., 2006). Thereengagement subscales eachwith three items: vigor (e.g.,k, I feel bursting with energy), = .73; dedication (e.g.,siastic about my work), = .74; and absorption (e.g., Ied in my work), = .64. We added all items together toposite scale. Items were measured on a 7-point Likertng from 1 (never) to 7 (every day); = .84.

    nalysis

    r to assess common method variance, followingndOrgan (1986),weconductedHarmans one factor testcale items were entered into an unrotated factor anal-ctors with eigenvalues greater than 1 were extracted;counted for 24.39% of the variance, and the next nineor 33.83%, indicating that not a single factor, nor a gen-ccounted for the majority of the covariance among thet and criterion variables. Common method bias doesto be a serious problem in the study.e multilevel nature of the data, we used hierarchi-odeling (HLM) and the computer program HLM 6.06h et al., 2005) to test all hypotheses. We standardizedables across individuals, and standardized Level 2 vari-the 30 organizations (Mathieu and Taylor, 2007) to: (a)comparison of measures that use different scales; andwith the graphical interpretation of interactions.

  • Please cite a lead indicator of workplace bullying and harassment, jobresources, 11), doi:10.1016/j.aap.2011.04.010

    ARTICLE IN PRESSGModelAAP-2439; No.of Pages126 R. Law et al. / Accident Analysis and Prevention xxx (2011) xxxxxx

    To test formeso-mediation,which is concernedwith cross-levelprediction of intercepts, we followed Mathieu and Taylors (2007)recommended steps. Preliminary steps required an assessment ofthe null model and relative magnitude and signicance of variance(ANOVA) thtion) for eac

    To testcross-level(Table 2). S(Table 3). Stthat persistno signicaof these stelevel variablevel measumally usingmediation (vals of thewere group

    To test mtion, controin the slopetionship by

    Level 1 : Ps

    Level 2 :B

    Hypothecome.

    2.5. Aggreg

    To asseseyes of theor group leby assessinintra-classorganizatioindex (Jameity coefcieaccordinglyalpha (inter

    One-washowed sig.1185, indicsafety climazations. Anresearch (Ja(SD= .18), >chosocial sa2000). We asistenciesuthis was grethese resultprovide suption.

    3. Results

    3.1. Data sc

    Prior tofor accuracdistributionmultivariat Ta

    ble

    1M

    eans,

    stan

    dar

    ddev

    iation

    san

    dco

    rrel

    atio

    nsbe

    twee

    nm

    odel

    variab

    les.

    MSD

    12

    34

    56

    78

    910

    1112

    1314

    15F

    ICC

    (1)

    1.Age

    41.0

    212

    .89

    .19

    .32

    .15

    .14

    .09

    .02

    .15

    .20

    .02

    .42

    *.

    17.

    43*

    .18

    .04

    2.02

    **.1

    925

    2.Gen

    der

    .17

    *.

    50**

    .35

    .43*

    .21

    .32

    .52*

    *.

    20.0

    1.0

    8.

    21.2

    0.2

    4.0

    63.

    40**

    *.2

    688

    3.In

    com

    e5.

    442.

    26.2

    8**

    .36

    **.

    25.0

    5.1

    1.

    63**

    .10

    .39*

    .48*

    *.0

    1.4

    1*.

    33.

    20.3

    13.

    93**

    *.3

    390

    4.Bullyi

    ng

    and

    har

    assm

    ent

    0.0

    31.

    49.0

    1.1

    6**

    .00

    .26

    .30

    .45*

    .45*

    .59

    **.

    31.

    02.

    58**

    .58*

    *.2

    9.

    201.

    81*

    .136

    45.

    Wor

    kpre

    ssure

    5.90

    2.95

    .01

    .17*

    .28*

    *.2

    5**

    .02

    .16

    .70*

    *.

    09.

    01.

    18.

    13.2

    0.1

    6.1

    91.

    33.0

    908

    6.W

    ork

    hou

    rs32

    .36

    16.4

    8.1

    7*.

    16*

    .36*

    *.1

    2.0

    9.

    01.1

    6.

    10.

    03.0

    5.

    09.3

    7*.0

    3.2

    02.

    75**

    *.1

    938

    7.Ph

    ysic

    aldem

    ands

    10.4

    23.

    21.

    12.0

    7.

    30**

    .14*

    .15*

    .08

    .11

    .36

    *.

    27.

    17.

    48**

    .26

    .06

    .13

    4.62

    ***

    .519

    58.

    Emot

    ional

    dem

    ands

    10.4

    62.

    68.1

    9**

    .28*

    *.1

    9**

    .27*

    *.5

    5**

    .20

    **.1

    8**

    .09

    .07

    .03

    .38

    *.2

    8.3

    0.1

    44.

    28**

    *.3

    910

    9.Pr

    oced

    ura

    ljust

    ice

    11.1

    61.

    75.

    00.

    12.1

    1.

    28**

    .28

    **.

    05.

    13.

    20**

    .44*

    .03

    .56*

    *.

    48**

    .22

    .32

    2.02

    **.1

    063

    10.O

    rgan

    izat

    ional

    rew

    ards

    6.44

    1.77

    .08

    .03

    .23*

    *.

    14**

    .17

    *.0

    8.

    24**

    .09

    .27*

    *.0

    1.5

    4**

    .33

    .18

    .50*

    *1.

    28.0

    662

    11.S

    uper

    viso

    rsu

    ppor

    t9.

    511.

    56.

    10.

    03.

    04.

    27**

    .28

    **.

    02.

    13.

    14*

    .29*

    *.1

    9**

    .37*

    .19

    .05

    .13

    2.21

    **.2

    296

    12.P

    sych

    osoc

    ials

    afet

    ycl

    imat

    e41

    .79

    10.3

    7.

    15*

    .08

    .02

    .38

    **.

    38**

    .02

    .35

    **.

    39**

    .61*

    *.4

    2**

    .40*

    *.

    21.

    29.4

    7**

    1.74

    *.1

    185

    13.P

    sych

    olog

    ical

    distr

    ess

    15.3

    04.

    65.

    15*

    .18*

    .04

    .39*

    *.3

    3**

    .06

    .18*

    *.2

    8**

    .22

    **.

    29**

    .18

    **.

    26**

    .40*

    .18

    1.11

    .032

    414

    .Em

    otio

    nal

    exhau

    stio

    n16

    .27

    7.55

    .15

    *.1

    3.1

    3.3

    8**

    .51*

    *.1

    1.1

    2.3

    6**

    .23

    **.

    22**

    .15

    *.

    32**

    .57*

    *.

    660.

    91.0

    316

    15.E

    nga

    gem

    ent

    50.4

    49.

    85.1

    4*.0

    0.1

    5*.

    18**

    .10

    .13

    .21

    **.0

    7.2

    2**

    .38*

    *.1

    5*.4

    2**

    .34

    **.

    43**

    2.41

    ***

    .161

    6

    Not

    e:*p