International Journal of Nursing Studies 44 (2007) 297–314 Literature review Impact of job satisfaction components on intent to leave and turnover for hospital-based nurses: A review of the research literature Billie Coomber, K. Louise Barriball King’s College London, London, UK Received 12 October 2005; received in revised form 31 January 2006; accepted 10 February 2006 Abstract Background: The United Kingdom (UK), alongside other industrialised countries, is experiencing a shortage of nurses partly due to low retention rates. Job satisfaction has been highlighted as a contributing factor to intent to leave and turnover, yet this is a complex area with many elements affecting its measurement. Aim: The aim of this paper is to explore the impact of job satisfaction components on intent to leave and turnover for hospital-based nurses in order to identify the most influential factors. Methods: To achieve this, a systematic search of the literature was undertaken to identify relevant international research. Three databases (i.e. BNI, CINAHL and PsychInfo) were utilised, resulting in nine articles that met the inclusion criteria. Four recurrent themes were identified in the literature: leadership, educational attainment, pay and stress. Findings: The key findings suggest that stress and leadership issues continue to exert influence on dissatisfaction and turnover for nurses. Level of education achieved and pay were found to be associated with job satisfaction, although the results for these factors were not consistent. Conclusion: Investigating possible changes over time in sources of dissatisfaction revealed that factors related to the work environment rather than individual or demographic factors were still of most importance to nurses’ turnover intentions. The differences found to occur across work settings necessitates analysis of job satisfaction at ward level, and the contribution of qualitative methods to develop more detailed insight is emphasised. The inconsistent findings over time associated with the effects of educational attainment and pay on intent to leave suggest that it is imperative that sources of job satisfaction are reassessed in the light of ongoing changes. r 2006 Elsevier Ltd. All rights reserved. Keywords: Job satisfaction; Nurse turnover; Intent to leave; Nurse retention; Nurse shortage; Attrition; Research review What is already known ?  Recruitment and retention of nurses is a contemporary issue of concern internationally as well as in Britain.  Job satisfaction is a concept closely linked to intent to leave and, hence, turnover within the nursing profession.  A multitude of factors can exert an effect on the job satisfaction of nurses. ARTICLE IN PRESS www.elsevier.com/locate/ijnurstu 0020-7489/$-see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2006.02.004 Corresponding author. Tel.: +442072740348. E-mail address: [email protected] (B. Coomber). What this paper adds  Reviews international research on job satisfaction exposing the factors most influential on turnover.  Highlights the need to analyse job satisfaction at individual ward level.  Considers the contribution of qualitative research methods to develop insight.  Reconfirms organisational factors (particularly stress and leadership) as more influential than individual factors (or demographics) on job satisfaction and turnover intentions. 1. Introduction For 2003, the Office of Manpower Economics (OME) (2003) reported that the rate of nursing staff leaving the National Health Service (NHS) in Britain was 9.4%. Given that nurses account for approximately 356,000 members of staff in the NHS, the scope of this ‘wastage’ amounts to approximately 33,500 nurses per annum. Nursing shortages have effects on many aspects of health care in Britain, not least constraining the effective delivery of services (Department of Health (DH), 2001). The problems associated with the shortage of registered nurses are compounded by an increasing demand for health care due to advances in medical technology, longer life expectancy, and the subsequent number of people living with serious illness and chronic disease (Buchan and Seccombe, 2003). Furthermore, there are increased demands for nurses as roles have widened to include those responsibilities formerly undertaken by medical staff, which emphasises the negative impact a nursing shortage may have on the quality of patient care (Aiken et al., 2002). A fundamental policy shift has occurred in the British NHS with the introduction of modernisation plans and the implementation of strategies for improving the working lives of health care staff (DH, 2000, 2001). The NHS Plan (DH, 2000) acknowledges workforce planning and effective approaches to recruitment, retention and motivation of NHS nurses as key elements to achieving a modernised health service. However, the health gain targets imposed by the NHS Plan are dependent on increasing the numbers of nurses in the NHS (Buchan and Seccombe, 2003). 2. Background 2.1. Turnover Turnover at an organisational level is seen to be a major contributor to the shortage of nurses (Gauci- Borda and Norman, 1997). Additionally, the loss of performance and efficiency on part of the leaver prior to departure is a major consequence of turnover, and high absence is seen as a critical problem in times of high turnover (Cavanagh, 1989). This has the effect of increasing pressure on, and decreasing morale of the remaining staff, resulting in the possibility of further turnover (Gauci-Borda and Norman, 1997). The im- plications of this ‘vicious cycle’ serve to focus attention on retention as a means to inhibit turnover and address the burden of shortages. There are inconsistencies within the literature regard- ing turnover, which may be attributable to the uncertainty surrounding both definition and measure- ment (Cavanagh, 1989). Turnover is generally viewed as the movement of staff out of an organisation, yet this consistently presents problems within turnover research for two reasons. Firstly, the incongruent inclusion in sample populations of staff who are retiring or are dismissed alongside those who voluntarily leave; and secondly, distinction between intra- (i.e. movement between units within the same organisation) and extra- (i.e. movement across organisations) institutional turn- over is not made freely within the literature, which has particular resonance for the health service. Bluedorn (1978) viewed turnover as a two-dimensional concept, distinguishing between the act of leaving as voluntary or involuntary, and between the leaving and joining of an individual to an organisation. This defined view is useful to consider when attempting to distinguish the variables that may have a relationship with voluntary turnover. To promote clarity, the theorists Fishbein and Ajzen (1975) sought to explain factors that predict actual turnover, summarising that behavioural intention is the primary antecedent to actual behaviour. This infers that the cognitive process of turnover intention (intent to leave) is an important predictor of actual turnover, a concept that has much empirical and theoretical support within turnover research (Mobley et al., 1978; Bluedorn, 1982; Steele and Ovalle, 1984; Prestholdt et al., 1987). Therefore, turnover intention (or intent to leave/stay) is consequently considered as an outcome of affective variables (such as job satisfaction) rather than actual turnover, which may be mediated by other variables such as age and tenure (Hellman, 1997). Actual turnover can thus be tested, but may not be subject to variables that are directly predictive. 2.2. Job satisfaction As a shortage ensues and difficulties in retention are highlighted, it follows that the reasons nurses leave their jobs must be clearly identified if the issue is to be successfully addressed. Job satisfaction has been cited as a major contributory factor to intent to stay in the international nursing literature (Cavanagh, 1992; ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314298 Gauci-Borda and Norman, 1997; Strachota et al., 2003), but job satisfaction is a complex phenomenon with many affecting components. Its status as an important predictor of intent to stay has the secondary effect of decreasing turnover, with many authors concluding that a decrease in turnover occurs when a workforce is satisfied (Saleh et al., 1965; Blegen, 1993; Irvine and Evans, 1995; Hellman, 1997). Whilst substantial literature exists regarding job satisfaction among employees in general and within nursing specifically, there appears to be no agreed precise definition (Cavanagh, 1992). It has been de- scribed as an individual attitude to how well personal expectations at work correspond to outcomes (McKen- na, 2000). Therefore, an individual’s appraisal of the degree to which the job fulfils one’s own job values can cause a positive emotional state of satisfaction or a contrasting negative feeling of dissatisfaction. Job satisfaction is considered within empirical studies either as an overall (global) feeling about the job, or as a related set of attitudes about various aspects of the job (facet approach) (Spector, 1997). The global approach to measurement is used when the interest is in overall attitude to the job. However, facet approaches can determine which particular aspects of the job are producing satisfaction or dissatisfaction for the individual and are, therefore, important in determining areas for improvement. Facets of job satisfaction can involve any aspect of the job and those frequently assessed include pay, co-workers, super- visors, organisational factors and work environment (Smith et al., 1969; Stamps and Piedmonte, 1986). 2.3. Concept of job satisfaction Whilst definitions can provide a broad understanding of what job satisfaction entails, it remains a complex concept illustrated by the multiple variables that have been studied in relation to it. Different theories present differing conceptualisations of job satisfaction that can be categorised as ‘content’ or ‘process’ theories (Campbell et al., 1970). Herzberg’s (1966) Motivator– Hygiene theory and Maslow’s (1970) Hierarchy of Needs exemplify content theories, and attempt to identify needs or values to be realised in order for an individual to be satisfied at work. For instance, Herzberg (1966) identified maintenance and motivating factors related to peoples’ attitudes to work: main- tenance or ‘dissatisfiers’ such as pay and associated benefits, organisational policies and working environ- ment; and motivating factors including recognition, achievement and self-satisfaction. Together with defini- tions, two main themes run through these theories: that job satisfaction has an affective component, i.e. a feeling of satisfaction, and a perceptual component which is an evaluation of whether one’s job meets one’s needs (Tovey and Adams, 1999). This is particularly pertinent when appraising the reasons why different studies utilise differing measurement approaches, and provides ratio- nale for this occurrence. 2.4. The link between job satisfaction and turnover According to Hellman (1997), increasing dissatisfac- tion in employees results in a higher chance of considering other employment opportunities. In his meta-analysis of US studies of non-nursing workers, the relationship between job satisfaction and intent to leave was found to be significantly different from zero and consistently negative. These findings were reiterated in nursing research with many authors concluding that increasing job satisfaction decreased rates of turnover (Saleh et al., 1965; Price and Mueller, 1981; Cavanagh and Coffin, 1992). Seccombe and Smith (1997) found that the factors given by nurses as reasons for leaving were centred on issues known to affect job satisfaction such as ineffective supervisory relationships and poor opportunities for professional development, rather than external labour market forces of which managers would justifiably feel unable to control. A summary of disciplinary perspectives contributing to the understanding of nurse turnover behaviour was provided by Mueller and Price (1990). This included: economic research with its emphasis on individual choice and labour market variables; socio- logical research emphasising characteristics of the work environment and content; and psychological research which emphasised individual variables and cognitive processes. Irvine and Evans (1995) developed this concept into a model for their meta-analytical study on job satisfaction and turnover, a summary of which can be seen in Fig. 1. 3. Methods 3.1. Search strategy The aim of this review was to explore the impact of job satisfaction components on intent to leave and turnover for hospital-based nurses. In order to extract relevant research from the published literature to achieve this aim the electronic databases Cumulative Index to Nursing and Allied Health Literature (CI- NAHL 1982–2004), Psychology Information (PsycIN- FO 2000–2004) and British Nursing Index (BNI 1985–2004) were searched. Keywords were ‘nurses’, ‘retention’ and ‘job satisfaction’ with synonyms and phrases being used as appropriate (for example, the term retention was used and combined with attrition, intent to leave, intent to quit, propensity to leave, intent to stay and turnover). The years 1997–2004 were chosen as a limit option in order to select only recent published work ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 299 that may hold more relevance for the nursing profession today due to the rapidly changing nature of nursing over recent years. Where possible, the search was globally limited to research (e.g. within CINAHL) and manually where electronic search limits were not possible. Specific criteria for inclusion and exclusion (Table 1) were also applied to ensure fulfilment of the aim of this review and manual scanning of abstracts was undertaken for the purpose of checking article suitability. While it was not possible to retrieve a small number of papers, the application of the inclusion/exclusion criteria resulted in nine research articles being selected. These articles are summarised in Table 2. 3.2. Framework for analysis With a plethora of themes found within the chosen literature, a decision was made to systematically identify the most frequently cited using an adaptation of thematic content analysis. This entailed scrutinising each paper, logging each theme contained within each paper and selecting those themes with the greatest counts for analysis. The four themes with the highest frequencies were: (1) leadership, (2) educational attain- ment, (3) stress and (4) pay. Whilst an alternative approach to inclusion could have been to choose the most statistically significant themes, this was not ARTICLE IN PRESS ECONOMIC FACTORS • • • Pay Job market Training JOB SATISFACTION BEHAVIOURAL INTENTIONS STRUCTURAL FACTORS Work environment Work context TURNOVER PSYCHOLOGICAL FACTORS Individual Demographic • • • • Fig. 1. Model of nurse turnover behaviour (Irvine and Evans, 1995). Table 1 Showing criteria for inclusion and exclusion of articles Inclusion Exclusion Adult nurses Paediatric/mental health/learning difficulties nurses Hospital nursing Primary/community/correctional/public health/hospice/nursing homes/NHS direct/military Nurses in general medical and surgical units Solely specialist areas e.g. ICU, oncology, theatres, nurse practitioners, management, nurse anaesthetists Both job satisfaction+intent to leave/turnover a key focus of the paper Only job satisfaction covered Primary or secondary research Literature reviews B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314300 possible due to the difficulties in comparability of the data and the variation in reports of each statistic. 4. Findings 4.1. Leadership One of the problematic issues regarding leadership is the lack of consensus regarding definition. Despite this, some recognition that it refers to either the direct superior or ward manager exists and that certain characteristics pertaining to leadership are particularly pertinent in influencing intent to leave. Five of the nine papers used scale measurement of leadership styles and characteristics used in ward environments (Fang, 2001; Fletcher, 2001; Tzeng, 2002; Yin and Yang, 2002; Larrabee et al., 2003). Whilst leadership was defined differently in each study, similar results were obtained with one exception. Fang (2001) used a specifically constructed ‘super- vision satisfaction’ scale consisting of five items measur- ing competency, fairness, recognition, communication and collaboration (strongly disagree ¼ 1, strongly agree ¼ 5). Regression results indicated that supervisor satisfaction was one of the most significant predictors of turnover intention (coefficient standardised ¼0.168, po0:05) but not turnover cognition (0.032). The study surveyed 180 nurses in Singapore and obtained a 90% response rate strengthening the validity of the results although it was undertaken in a single hospital, thus limiting generalisability. Fletcher (2001) used a six-item scale of ‘immediate supervisor’ measuring perceived qualities such as supervisor reliability, competency and helpfulness (very true ¼ 1, not at all true ¼ 4), as well as the supervisory subscale from the Job Diagnostic Survey (JDS, Hackman and Oldham, 1975) (extremely dissatisfied ¼ 1, extremely satisfied ¼ 7). The definition given for immediate super- visor was ‘generally the nurse manager’, and the lowest ratings attained were related to this subject. The lowest individual item means for the subscale of JDS were for supervisor support (mean ¼ 4.34) and quality of super- vision (mean ¼ 4.46). For the Immediate Supervisor scale the highest means (and therefore lowest ratings) were for supervisor helpfulness (2.81) and reliability (2.68). The low ratings were clarified with written comments from participants that were categorised into four general themes: quality of leadership, lack of physical presence, failure to address problems and staffing issues. Whilst the results pertaining to intent to leave indicated that the participants surveyed were likely to remain in nursing, this was measured with a single question scored on a 5-point Likert scale without further statistical tests applied. Despite the large sample size (n ¼ 1780 for JDS, n ¼ 509 for qualitative component) and the number of hospitals surveyed (n ¼ 10), the response rate for both components was low (34.5% and 28.6%). Furthermore, no report or profile of non- respondents was provided, which may have affected the representativeness of this US sample. Larrabee et al. (2003) also reported associations between leadership and intent to leave, but acknowl- edged that this was an indirect link. Their US study collected data (n ¼ 90, RR ¼ 60%) via nine subscales of the Multifactor Leadership Scale (Bass and Avolio, 1995) with varying numbers of items in each subscale: transformational leadership (5), transactional leadership (3), non-transactional (1) and leadership outcomes (3). Intent to leave was measured using a single item (definitely will not leave ¼ 1, definitely will leave ¼ 5). The results of stepwise multiple regression suggested that leadership was not directly related to intent to leave but indirectly related through various other predictors. Transformational leadership style was found to predict psychological empowerment of nurses (po0:001) which in turn predicted an increase in job satisfaction (po0:001) which was then a major predictor of a decrease in intent to leave (po0:001). These findings illustrate the complexity involved in various perceptual and attitudinal factors contributing to an individual’s intention to leave. In contrast to the findings of the studies presented above, the results of Tzeng’s (2002) work suggested that leadership style exerted no influence on job satisfaction and intent to leave (n ¼ 648, RR ¼ 82%). The different method of modelling utilised by Tzeng distinguishes this study from the others and may account for the variance in findings, although it is noteworthy that large p-values (model 2, p ¼ 0:10; model 3, p ¼ 0:35) were obtained for leadership style. This study was undertaken in Taiwan and whilst the potential for cultural influences to explain the differences is acknowledged, interestingly Yin and Yang (2002) attained opposing results in their meta- analysis of 13 Taiwanese studies. In particular, consis- tently significant results were reported for experienced turnover and potential turnover due to supervisor dissatisfaction (ranked 5/18 and 3/16, respectively, po0:05). The specific areas of supervisor dissatisfaction, while measured, were not presented thus lowering the opportunity for direct application of findings to practice. 4.2. Educational attainment Conflicting results were found for the effect of the individual factor of educational attainment on job satisfaction and on intent to leave. Six studies used self-reported demographic data to ascertain the educa- tional level of participants (Fang, 2001; Lu et al., 2002; Tzeng, 2002; Yin and Yang, 2002; Larrabee et al., 2003; Rambur et al., 2003). Three studies found positive correlations for educational level and job satisfaction. A ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 301 ARTICLE IN PRESS Table2 Summarisingliteratureincludedforanalysis TitleandAuthor(s)ResearchquestionMethod,sample&datacollectionDataanalysisFindings Theeffectsofnurses’jobsatisfactionon retention:anAustralianperspective CowinL(2002) JournalofNursingAdministration32(5): 283–291 Anexplorationofthejobsatisfactionand retentionplansofnewlygraduatedand experiencednursesinNSWandmeasures theintentionofnursestostayinnursing Descriptive,correlationaldesignwith qualitativecomponent.Survey questionnaireadministeredon-siteto2 groupsofdifferingexperienceinNSW. Longitudinalmeasurementofjob satisfaction:samegroupsmeasuredat0 monthand8months: Gp.1:studentRNsinlastsemesterofBN at6universities 0monthsðT1Þ8monthsðT2Þ RR¼58%RR¼22% n¼506n¼110 Gp.2:randomselectionofRNsfrom NSWNurses’RegistrationBoarddatabase RR¼26:4%RR¼63% n¼529n¼323 IWS Scoresfrom2partsofIWScombinedto provideoveralltotalweightedscore(the index).Pairedcomparisonsproducedrank ofimportanceof6factors.Rankof satisfactiondeterminedbymeanscorefor eachfactor. NRI Statisticalanalysisnotspecified Qualitativedata Wordprocessingprogrammeforinitial clusteringofthemes,thencontentanalysis. Thegreaterthesatisfactionwith professionalstatusthemorelikelythe intenttostayinnursing Jobsatisfactionremainedstableover timeforexperiencednurses. Paysignificantareaofdissatisfaction overtimefornewgraduates. Qualitativedata Issuesraised: burnout excessivestress perceptionofpoorpublicimage highresponsibilityforpoorpay. Questionnaireincluded: TheIndexofWorkSatisfaction (IWS). Nurses’RetentionIndex(NRI) (designedbyauthorandadministered tobothgroupsatT2). Blankpageforcomments. Turnoverpropensityanditscausesamong Singaporenurses:anempiricalstudy FangY(2001) InternationalJournalofHumanResource Management12(5):859–871 Examinationofvariousantecedentsof turnoverattemptingtoidentifymost criticalcausesoftheturnoverproblemin Singapore.Totestpreviousresearch frameworksontheirexternalvalidity acrossnations. Surveyquestionnaireadministeredto randomsampleofRNsviaon-sitesurvey sessionsatalargegeneralhospitalin Singaporeincluding: Demographics Nine-itemstressscale JobDescriptiveIndex(JDI) OrganisationalCommitmentscale (modified) Supervisionsatisfactionscale(created byauthor) Professionalcommitmentscale (createdbyauthor) Turnovercognitionandturnover intentioneachmeasuredwithone question. Descriptivestatistics,correlationand multipleregressionproceduresused.Stressthroughoutdifferentstagesin turnoverdevelopment. Professionalcommitmentasignificant causeofturnovercognitionbutnot turnoverintention. Organisationalcommitmentand supervisionsatisfactionamongthetop predictorsofturnoverintention. RR¼90% n¼180 B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314302 ARTICLE IN PRESS HospitalRNs’jobsatisfactionsand dissatisfactions (datausedsuppliedfromlargerstudy addressingwork-relatedstressinRNsand itsrelationtoillnessandinjury) FletcherC(2001) JournalofNursingAdministration31(6): 324–331 Largerstudy AnexaminationofwhetherstressinRNs isassociatedwithillnessorinjury. PostalquestionnairesenttoRNsin10 hospitalsinsouthernMichiganincluding: TheSpecificSatisfactionsSubscale (fromJobDiagnosticSurvey) TheImmediateSupervisor TheHealthProfessionsStress Inventory Singlequestionrelatingtointentto leave. Meanofscorescalculatedforeachitemon scales. Furtherstatisticaltestsnotdescribed. Responsestoquestionnaire: RNsslightlysatisfiedwithjobs ‘True’to‘somewhattrue’that supervisorshadqualitiessuchas reliability,competency,helpfulness Jobssometimestosomeextent stressful RNslikelytoremaininnursing. RR¼34.5% n¼1780 Smallerpartofstudy Anexplorationofjobsatisfactionfactors (includingturnover)relatedtothework environment. Qualitativecomponent Blankpageandinviteforcomments reworkinginhospitaladministeredwith above. RR¼28:6%ðof1780Þ n¼509 Thematicanalysisanddescriptionof commentsprovided.Contentanalysisnot disclosed. Qualitativedataissues: Jobperformance Intrinsicworkvalues Patientcareissues UnlicensedAssistivePersonnel Jobsecurity,salaryandfringe benefits,workschedules QualityofLeadership. Predictingregisterednursejobsatisfaction andintenttoleave LarrabeeJ,JanneyM,OstrowC,Withrow M,HobbsG,BurantC(2003) JournalofNursingAdministration33(5): 271–283 Investigationoftherelativeinfluenceof nurseattitudes,contextofcareand structureofcareonjobsatisfactionand intenttoleave. ConveniencesampleofRNson2medical, 2surgicaland3intensivestep-downunits inuniversitymedicalcentreinWest Virginia,USA. On-sitequestionnaireincluding: WorkQualityIndex NinesubscalesofMultifactor LeadershipQuestionnaire NurseCollaborativePracticeScale SupportServicesInstrument GroupCohesionScale PersonalViewsSurveyIII IntenttoLeavequestion(5-point responsescale) PsychologicalEmpowerment12-item scaleDemographicdata. RR¼90% n¼90 Dataonunitturbulenceandstaffing collectedviaexistingwardrecords. Analysisofvarianceevaluateddifferences inintenttoleaveandjobsatisfactionon thebasisofthecategoricalvariablesto identifyanythatrequiredstatistical controlduringmultivariateanalysis. Predictorsofintenttoleavefirstevaluated usinglogisticregression,thenstepwise multipleregression Majorpredictorofintentto leave¼jobdissatisfaction. Majorpredictorofjob satisfaction¼psychological empowerment. Predictorsofpsychological empowerment¼hardiness, transformationalleadershipstyle, nurse/physiciancollaboration,group cohesion. B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 303 ARTICLE IN PRESS Therelationshipsamongturnover intentions,professionalcommitmentand jobsatisfactionofhospitalnurses LuK-Y,LinP-L,WuC-M,HsiehY-L, ChangY-Y(2002) JournalofProfessionalNursing18(4): 214–219 Aninvestigationoftherelationships amongturnoverintentions,professional commitmentandjobsatisfactionofRNs. Postalquestionnairesenttorandom sampleofhospitalnursesinsouthern Taiwan(addressesprovidedby3Taiwan Nurses’Associations)including: ProfessionalCommitmentScale 2questionsre:intenttoleave(4point responsescale) Jobsatisfaction4pointscaleof10 items(developedbyauthors). Descriptivestatisticsusedtoexplore variables. Correlationmatricesgeneratedtoexamine relationshipsbetweenvariables. Discriminantanalysesusedtoestimate% ofcorrectclassificationsofintenttoleave onjobsatisfactionandprofessional commitment. +vecorrelationbetweenjobsatisfaction and: professionalcommitment intenttoleaveorganisation intenttoleaveprofession. vecorrelationbetweenjobsatisfaction andturnoverintentions. Educationallevelsignificantlyrelatedto jobsatisfactionandturnover. AStatewideanalysisofRNs’intentionto leavetheirposition RamburB,PalumboM,McIntoshB, MongeonJ(2003) NursingOutlook51(4):182–188 Whateffectdoesgender,age,educational preparation,setting,position,clinical practiceareaandpopulationdensityhave onnurses’intentiontoleavetheircurrent primarypositioninthenext12monthsfor careeradvancement,situationalorjob dissatisfactionreasons? SecondaryanalysisofRNrelicensure surveys(self-report)returnedin2001in Vermont,USA. Questionsincludedontopicsof: Intenttoleave Reasonforleavingcurrentposition Practicesetting Enrollmentineducationalprogram Demographics. RR¼80% n¼4418 Associationtestingusedw2forcategorical variablesandt-testsforcontinuous variables. Stepwiselogisticalregressionusedto predictlikelinesstoleave. Allformalstatisticaltestinguseda0.05 typeIerrorlevel Differencesinintenttoleavevaryby educationalattainment,hoursworkedper week,gender,practiceroleandpractice activity. Intenttoleavepositionforcareer advancementreasonsincreasedwith levelofeducation. Intenttoleaveforreasonsassociated withjobdissatisfactiondecreasedwith educationallevel. Intenttoleaveforreasonsassociated withjobdissatisfactionwashigher thanleavingforreasonsassociated withcareeradvancementor situationalreasonsacrossall educationalpreparations. Nostatisticallysignificantoverall genderdifferencesinintentionto leave. Thosecurrentlyenrolledin educationalprogrammeslesslikelyto leave. Nursesaged51oroverlesslikelyto leaveforreasonsofjobdissatisfaction comparedtonursesin40–50age group. FactorsInfluencingSatisfactionand anticipatedturnoverfornursesinan academicmedicalcenter Anexaminationoftherelationships betweenworksatisfaction,stress,age, Cross-sectionalsurveydesign. On-sitequestionnaireadministeredto241 staffnursesand5nursemanagersfrom12 Pearson’scorrelationanalysesperformed forallmajorstudyvariables. mjobstress¼kgroupcohesion. kjobsatisfaction. manticipatedturnover. Table2(continued) TitleandAuthor(s)ResearchquestionMethod,sample&datacollectionDataanalysisFindings B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314304 ARTICLE IN PRESS ShaderK,BroomeM,BroomeC,WestM, NashM(2001) JournalofNursingAdministration31(4): 210–216 cohesion,workscheduleandanticipated turnoverinanacademicmedicalcentre. generalmedical-surgicalandICUina universityhospitalinsoutheasternUSA including: TheIndexofWorkSatisfaction JobStressScale(modified) BryneGroupCohesionScale AnticipatedTurnoverScale Demographicdata. StaffnursesManagers RR¼63%RR¼42% n¼151n¼2 Unitdemographicquestionnairecollected informationonunitsize,turnoverdata, patientsatisfactionscores,overtime, acuity,averagedailycensus,staffingmix andreallocation. Morestable workschedule¼kwork-relatedstress. kanticipatedturnover. mgroupcohesion. mworksatisfaction. Theresultsshoweddifferencesin significantpredictorsofanticipated turnoverfordifferingagegroups: 20–30yrs¼jobsatisfactionandjobstress 31–40yrs¼jobsatisfaction 41–50yrs¼jobsatisfactionandgroup cohesion 51andover¼nosignificantpredictors. Theinfluenceofnurses’working motivationandjobsatisfactionon intentiontoquit:anempirical investigationinTaiwan TzengH-M(2002) InternationalJournalofNursingStudies 39:867–878 Hypothesis:thehighernurses’levelson generaljobsatisfaction,overall satisfactionwithprofessionalroles,and generaljobhappiness,thelowerthe intentiontoquitwouldbe(after controllingfordemographic characteristicsandworkingmotivation). Questionnairesdistributedtonursesin3 hospitalslocatedinthesamedemographic areaofsouthernTaiwan.9jobsatisfaction scalesused: Indirectworkingenvironment Directworkingenvironment Salaryandpromotion Self-growth Challengeinwork Interactionwithandfeedbackfrom patientsandfamilymembers Leadershipstyle Workingatmosphere Familysupportandreligion. 3generaljob-relatedindicators: Generaljobsatisfaction Overallsatisfactionwithprofessional roles Generaljobhappiness. RR¼82% n¼648 Ordinallogisticregressionanalysesusedto developamodelofnurses’intentionto quit. Generaljobsatisfaction,generaljob happiness,satisfactionwithsalaryand promotion,institution,educational backgroundandageofnurses’youngest childwerefoundtobesignificant predictorsofnurses’intentiontoquit. B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 305 ARTICLE IN PRESS NursingturnoverinTaiwan:ameta- analysisofrelatedfactors YinJ-C,YangK-P(2002) InternationalJournalofNursingStudies 39:573–58 Toprovideanorganisedoverallpictureof turnoveramongnursesinTaiwan. Meta-analysisof13studiesrelatedto nursingturnoverpublishedorreportedin theChineselanguageliterature.Inclusion criteriamaintainedthatstudiesmust reportanalysesofquantitativedatafrom samplesofhospitalnursesengagedin patientcare. Individualsamplesizesrangedfrom69to 725,withatotalof4032subjectsfrom109 hospitals. Dataintegratedbyestimationof parametriccorrelationcoefficients. DataanalysedusingFriedman’stwo-way analysisofvariancebyranks. Testofhomogeneityusedw2statistic. Differentlylabelledvariableswere combinedwhentheauthorjudgedthe variablestodefineconceptuallysimilar phenomena. Top10factorsrelatedtoexperienced turnover: Poorpromotionopportunities Workstressduetohighworkload Lackofcontinuingeducation Dissatisfactionwithsalary Dissatisfactionwithsuperior Inflexiblescheduling Administrativepolicies Dissatisfactionwithfringebenefits Recognition Unstablescheduling. Top10reasonsforpotentialturover: Salaryandfringebenefits Peergrouprelationships Leadershipstyleofdirectsuperiors Levelofchallenge Administrativepolicies Poorpromotionopportunities Workstressduetohighworkload Achievement Professionalism Workenvironment. Maritalstatusandeducationallevel correlatedwithjobsatisfaction. Educationallevelcorrelatedwith autonomy. Table2(continued) TitleandAuthor(s)ResearchquestionMethod,sample&datacollectionDataanalysisFindings B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314306 secondary analysis of US relicensure surveys by Rambur et al. (2003) (n ¼ 4418, RR ¼ 85%) reported that job dissatisfaction decreased with higher educational level, and Yin and Yang’s (2002) work in Taiwan confirmed this finding, suggesting that nurses with lower educa- tional levels are less satisfied with their jobs (po0:05). This is reiterated by Tzeng (2002), whose results for the final ordinal regression indicated that holding a diploma (estimate ¼1.790, Wald w2 ¼ 4:28, p ¼ 0:04), holding an associate degree (1.639, Wald w2 ¼ 4:00, p ¼ 0:05) and general job satisfaction (0.862, Wald w2 ¼ 22:89, p ¼ 0:00) are significant predictors of Taiwanese nurses intention to quit. Yin’s and Yang’s (2002) meta-analysis of individual factors related to nurse turnover asserted a small but stable relationship for educational attainment (rxy ¼ 0:09), with this relationship being one of the strongest out of the individual factors that included marital status, years of tenure, age and position. Interestingly, opposing results were obtained by Lu et al. (2002) who found a significant association indicating educational level to be negatively correlated with job satisfaction (Spearman’s correlation 0.101, po0:01) and positively correlated with intent to leave the organisation (.054, po0:05) and the profession (0.058, po0:01) for a Taiwanese sample (n ¼ 2197, RR ¼ 86.2%). The studies by Fang (2001) and Larrabee et al. (2003) both concluded that there were no significant relationships between any demographic factors and turnover cognition or intention. Fang (2001) further reported that job satisfaction did not exhibit any significant influence on turnover in a sample of Singaporean nurses (coefficients standardised for cognition 0.108; for intention 0.007). The small, non- random sample in Larrabee et al.’s (2003) study could indicate a reason for their opposing results, with larger studies required to confirm these as valid findings. Additionally, the purpose of their study did not specifically involve the investigation of demographic variables on job satisfaction and intent to leave, so results of statistical tests were not provided for educa- tional attainment, thus disallowing independent assess- ment of the effect. Overall, there was no consensus regarding educational attainment and its relationship with either job satisfac- tion or intent to leave. The inconsistent, yet sometimes significant results allow for considerable confusion when attempting to interpret the importance and implications of educational attainment for retention. Similarities in data collection methods and use of statistics across studies further hinders the clarification of reasons for the differences in findings. It could be assumed from these results that assessment of educational level is fraught with methodological challenges, or perhaps the samples were very distinct. Certainly the findings are mixed, and as such conclusions should be drawn with caution. 4.3. Pay Throughout the studies that measured the importance of pay for nurses (Fang, 2001; Fletcher, 2001; Cowin, 2002; Rambur, 2003), definitions were implied as ‘salary’ or ‘salary and fringe benefits’, which enhanced compar- ability of findings. However, the impact of salary related to culture is difficult to determine without detailed knowledge of the wider social and economic climate in which it operates. In addition, several different findings were produced from differing methods of assessment. Cowin (2002) measured satisfaction with pay over an 8-month period among a sample of Australian graduate nurses (n ¼ 506 at T1, n ¼ 110 at T2) and experienced nurses (n ¼ 528 at T1, n ¼ 332 at T2) through a component of the Index of Work Satisfaction (IWS, Stamps and Piedmonte, 1986) at two points in time (T1 ¼ 0 months, T2 ¼ 8 months). The findings indicated that although the issue of pay was not ranked as the most important, it rated as the least satisfying for graduates (T1 and T2) and second least satisfying for experienced nurses (T2). Furthermore, the importance of pay increased over time for both groups (from 3/6 to 2/6). This could be attributable to other confounding factors occurring over the 8 months of the study but not measured, or may indicate a growing concern that has the potential to become a considerable source of importance. However, the specifically constructed Nurse Retention Index (NRI, 8-point forced choice Likert) was administered to the two groups at T2, and Cowin (2002) stated that the results of multiple regression analysis suggest that pay was not a statistically significant indicator of intent to leave for either group (figures undisclosed). Despite this, the qualitative component of Cowin’s (2002) study, which consisted of written comments, indicated that dissonance between pay and level of responsibility may lead to retention issues. A major source of concern for graduates and experienced nurses was the perceived inequality of pay for high level of responsibility. In particular, experienced nurses compared their workloads, level of knowledge and responsibility to their perception of ‘less qualified’ professions, resulting in expressions of disappointment with career choice. Other comments suggested that pay was not a primary problem for retention when enjoy- ment of other aspects (undisclosed) of the work was high. Fang (2001) also found no statistically significant influence of pay on turnover cognition (coefficient standardised 0.108) or turnover intention (0.007). A complete assessment of the importance of pay was, however, complicated by the loading of pay into the same factor of job satisfaction. The overlap between these two factors was found during factor analysis and it was therefore found to possess a lack of empirical distinctiveness. The question asked of the sample ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 307 relating to pay also included the element of job security. It is possible that this multiple question could have affected the assessment of salary satisfaction if partici- pants were satisfied with the security of their position and scored accordingly. The inability to assess pay individually was also evident in Fletcher’s (2001) study. A job satisfaction subscale of the JDS (Hackman and Oldham, 1975) was used to measure pay satisfaction of US nurses (n ¼ 1780, RR ¼ 34.5%) via two items (Spearman–- Brown reliability for items on scale ¼ 0.86). The pay items were measured with a 7-point Likert scale (extremely dissatisfied ¼ 1, extremely satisfied ¼ 7) but the only result presented was that of an overall score for the subscale which consisted of 14 items. This asserted that participants were slightly satisfied with their jobs (mean ¼ 5.04, SD ¼ 0.99), but rendered specific results for the pay items inaccessible. Written comments from participants provided some clarity, with an expression of similar feelings to those from Cowin’s Australian (2002) study. They indicated feelings of inadequate reward for their ‘education, experience and expertise’, and unfair- ness of pay compared with other professions. Fletcher (2001) argued that the sample was likely to remain in nursing, although this conclusion was drawn on the basis of a single question providing little insight into the factors influencing participants’ responses. Generalisa- bility is compromised due to the low response rate (34.5%), and a reported demographic of nurses being represented by a ‘bargaining unit’ (type of union that negotiates salary each year) which is unrepresentative of the nursing profession in other countries. An interesting feature of the issue of pay and resulting influence for intent to leave for nurses was reported in Rambur et al’s (2003) large US study. A gender difference affecting pay satisfaction was found during secondary analysis of self-reported data. Pay was collapsed into the category of ‘job dissatisfaction’ along with aspects such as short staffing and poor manage- ment. Of those intending to leave (males ¼ 23%, females ¼ 20%), 75% of male participants were intend- ing to leave for reasons of job dissatisfaction in contrast to 51% of female participants (w2 ¼ 10:31, po0:01). In addition, male participants (53%) were more likely to be leaving due to dissatisfaction with salary than female participants (26%) (w2 ¼ 16:31, po0:01). As no statis- tically significant results for gender differences were obtained for the majority of other categories influencing intent to leave, gender expectations and socialisation may play a particular role in pay. 4.4. Stress Four studies specifically investigated the effect of stress on intent to leave, with stress consistently being cited as a major predictor of anticipated turnover. This had particular international significance with studies conducted in the US (Shader et al., 2001), Singapore (Fang, 2001), Australia (Cowin, 2002) and Taiwan (Yin and Yang, 2002) reporting similar results. Methods used to measure stress were varied and included scales (Fang, 2001; Shader et al., 2001), written comments (Fletcher, 2001; Cowin, 2002) and meta-analysis (Yin and Yang, 2002). Stress exerted the most substantial impact on turnover cognition (coefficent standardised ¼ 0.351, po0:000) and turnover intention (coefficient standardised ¼ 0.29, po0:001) of all variables measured in Fang’s (2001) study. Whilst reporting that stress explained 8.5% from a total of 41% of the variance for turnover cognition and 4.6% from a total of 32% for turnover, the specific causes of stress remained undifferentiated. This detracts from the wider focus of the study, which was to identify the most critical causes of the turnover problem in Singapore with a view to mitigating the problem. Whilst an indication that stress is a major cause is helpful for directing future research, a scale identifying specific stressors would have been helpful in ascertaining areas requiring improvement. Shader et al’s (2001) US study partly identified particular stressors via the Job Stress Scale (Hinshaw and Atwood, 1985), a 22-item, 4-point Likert with four subscales. The mean job stress score was 2.06 (SD ¼ 0.39, range 1–4), demonstrating that nurses reported moderate levels of job stress (n ¼ 151, RR ¼ 63%). The Anticipated Turnover Scale consisting of 12 items rated on a 5-point Likert measure (possible responses undisclosed) reflected individual perceptions about the possibility of leaving their current positions (mean score of 3.45, SD ¼ 1.06, range 1–7), rating their intention to leave as moderate. Pearson’s correlation analyses were performed on all major study variables. The main stressor was assessed to be lack of stability in the work schedule (constantly changing rota), with the more stable the schedule, the less work-related stress (r ¼ 0:205, po0:001) and lower anticipated turnover (r ¼ 0:29, po0:001). Higher job stress was also found to lower job satisfaction through the lowering of group cohesion (r ¼ 0:41, po0:001) which then increased anticipated turnover (r ¼ 0:37, po0:001). A stepwise regression model including the variables of job satisfac- tion (0.35), weekend overtime (0.27), job stress (0.16) and group cohesion (0.13) explained 31% of the variance in anticipated turnover. Shader et al. investi- gated turnover further by looking at different age groups in relation to these factors. The findings indicated that low job satisfaction and job stress were significant predictors of anticipated turnover for 20–30-year-old nurses (R2 ¼ 16, po0:001), low job satisfaction was a predictor for 31–40-year-old nurses (R2 ¼ 0:31, po0:001) and low job satisfaction together with low group cohesion for 41–50-year-old nurses (R2 ¼ 0:28, ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314308 po0:001). For nurses aged 51 and over, there were no significant predictors of anticipated turnover. The inference that stress is mainly a problem for the youngest nurses of the sample may be linked to the finding that the stressor of most importance was work schedule as high value may be placed on control over own time for this age group. The demographics denoted that two-thirds of the sample had been nursing for less than 3 years also highlighting the possible implication of inexperience. However, job dissatisfaction as measured by the IWS exerted an influence on intent to leave for all age groups of nurses implying ultimate significance of this factor for the study sample. A negative relationship between stress and job satisfaction (po0:01) was also reported as having an important influence on turnover in the meta-analysis of nursing turnover conducted by Yin and Yang (2001). Stress was inducted as one of 12 variables related to turnover from the factors included in 13 studies under- taken in Taiwan. The second most frequently reported reason for leaving last nursing position was work stress due to high workload. In contrast, stress was ranked 6th of reasons for potentially leaving a current job. This inconsistency in rank from experienced (2) to potential (6) turnover is difficult to explain, as logically it would be expected that factors would have to be important to intent to leave in order to influence turnover. This occurrence could possibly be explained by mediating factors that, perhaps, become less influential over time. For example, positive peer group relationships (ranked 2nd for potential turnover) within the workplace may buffer stress only to a certain point above which stress becomes more influential. However, another possibility may be that the retrospective measurement of reasons for turnover, with consequential reliance on partici- pants’ memory, could allow for inaccurate results. A strength of the meta-analysis makes the combined sample more representative of the population than individual samples; however, it is possible that the generalisability of these findings could be limited to hospital settings in Taiwan. Overall, for factors relating to turnover, stress has been consistently implicated in these studies, but ambiguity towards the actual stressors responsible has been shown. 5. Discussion 5.1. Limitations It is acknowledged that some of the studies selected for final inclusion contained data derived from some specialist areas due to random samples taken from the hospital workforce. Whilst measures were taken to reduce this occurrence, the reality of the empirical literature precludes this as a practicality. As a conse- quence, studies were included if the sample predomi- nantly contained nurses working in general medical or surgical environments. It is recognised that this may influence the conclusions drawn from the literature analysed. Another limitation of this review relates again to the selection of studies. Stricter criteria involving intent to leave could have ensured the review focused on intent to leave the profession rather than the organisa- tion. This may have wider significance for the profession than the organisation they serve. 5.2. Key outcomes and comparison over time In the meta-analyses of Blegen (1993) and Irvine and Evans (1995), factors with the strongest relationships to job satisfaction were found to be related to work content and environment. This remains true for more recent research findings as can be seen by the themes high- lighted. From the four themes discussed, three were organisational factors (leadership, stress and pay) and only one an individual/demographic factor (educational attainment). This has implications for improvements in retention as organisational changes may be more easily addressed than those involving age, education and tenure for nurses already employed. 5.3. Stress In earlier studies, stress had the strongest relationship with intent to leave, and factors such as low autonomy, low recognition and poor communication with collea- gues were described as the causes of this stress (Hinshaw et al., 1987; Blegen, 1993). With the changes affecting nursing practice over recent years and the corresponding increase in demands on nurses, it is unsurprising that stress is still an issue for today’s profession. It is unclear, however, what stage along the continuum of eustress to distress is being measured, and whether the amount of stress reported in recent work has increased or decreased from those of earlier studies. Additionally, specific stressors were not identified with any consensus in the papers studied although various stressors such as workload (Yin and Yang, 2001) and unstable work schedules (Shader et al., 2001) were identified as important for some samples. In an occupation as diverse and challenging as nursing, assessing stress is difficult (McVicar, 2003). This, coupled with the subjective nature of stress perception, renders the identification of specific con- tributing factors unrealistic. However, in the light of the consistent implication of stress influencing turnover, an effective reduction of stress is required to successfully address low retention. A full understanding of the stress phenomenon is necessary to facilitate appropriate policies and interventions, yet the fact that nursing ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 309 environments are not generic makes achieving this complicated. In view of this, interventions that seek to support the individual, such as stress-management training, which enables individuals to utilise their most effective coping strategies, and increasing support from colleagues and senior staff have both been suggested as appropriate measures to counteract stress within the nursing profession (McVicar, 2003). The UK Government recognises the need to address low retention in health care, with initiatives designed to reverse the trend in turnover. Introduction of flexible working practices are part of the Government’s commit- ment to improving the work–life balance of staff (DH, 1999). This may be particularly relevant for retention of younger nurses who were found to perceive stress from lack of control over work scheduling (Shader et al., 2001). An increase in staffing levels via improved retention is also a means to address the cited stressor of workload (Yin and Yang, 2001). With an increase in demand for health care, the burden of high workload can only be ameloriated by increasing nurse numbers. The RCN (2002) maintains this point, and adds that recruitment efforts will produce little significant im- provement in workload stress in the near future and should perhaps be seen as a medium to long-term measure. Government initiatives designed to meet needs of the majority is a starting point, but in view of the individual nature of stress may not provide the answer for all nurses (McVicar, 2003). In view of the fact that organisational change is a recognised source of stress (Kouzes and Posner, 2003) and that changes constantly occur in the health services in line with modernisation plans, this issue is unlikely to be solved imminently within the British NHS. 5.4. Leadership The variable of leadership, or supervisory relation- ship, has a consistent relationship with job satisfaction and intent to leave in both early and recent studies (Sorrentino, 1992; McNeese-Smith, 1995; Chiok Foong Loke, 2001; Fang, 2001; Fletcher, 2001; Yin and Yang, 2002; Larrabee et al., 2003). Nursing leadership style, or supervisory relationship, is a phenomenon of interna- tional relevance that is linked to the work environment. Fletcher’s (2001) study suggested that job dissatisfaction ensues when nurse managers fail to give due recognition and support, disregard staffing issues and neglect to address problems. Furthermore, Larrabee et al. (2003) asserted that the main effect on intent to leave was a decrease in job satisfaction through a lack of empower- ment perceived by nurses to be a result of leadership style. This suggests that when staffing levels are reduced, as they are for today’s profession, leadership that encompasses an open, empowering approach to staff may be able to buffer low retention. Transformational leadership style engenders motivation of others to pursue high standards, concentrates on creating open communication and a willingness to embrace change (Morrison et al., 1997; Upenieks, 2003). This would seem most desirable for a profession that experiences much change and that has expressed concern about supervisory relations to a point that it negatively affects turnover and possibly quality of care (McNeese-Smith, 1993; Davidson et al., 1997; Needleman and Buerhaus, 2003). Introduction of participative styles of manage- ment is central to the human resource proposals of the NHS Plan (DH, 2000), but change in leadership behaviour may not be easily obtained if this is not placed high on the agenda for managers (Finlayson et al., 2002). 5.5. Pay Nurses’ pay has been measured as a component on many scales, although early studies found it to contribute little to job satisfaction (Blegen and Mueller, 1987; Blegen, 1993; Irvine and Evans, 1995). However, the issue of pay and its contribution to job satisfaction increases in importance when nurses perceive discrepan- cies between their remuneration and that of other professions (Tovey and Adams, 1999). This was reiterated in the written comments obtained in the studies of Fletcher (2001) and Cowin (2002). It has also been suggested that perceived lack of reward contributes to role disengagement (Demerouti et al., 2000) and, although pay was not found to exert an effect on intent to leave in these studies, such perceptions have the potential to become influential. Comments from the participants of Cowin’s (2002) study suggesting that pay was not an issue only when other factors were satisfying reiterates this possibility. It is further supported by the dissatisfaction expressed by participants regarding the discrepancy between the high responsibility of the job and perceived low pay. As roles for nurses expand, and their responsibility increases, the potential for dissatisfaction and intent to leave could be amplified in the UK. Addressing pay via the Agenda for Change (DH, 2003) programme, which aims to modernise the salary system of the British NHS by providing competency- based pay (a variant of performance-related pay), has been suggested as a potential source of conflict for employees’ expectations, norms of fairness and practice (Grimshaw, 2000) although such a perspective has its detractors. In a case study undertaken as a precursor to Agenda for Change evaluation, Meerabeau et al.(2004) found that a competency-based pay system was valued by staff who felt that it provided structured role development. This dissonance demands that close attention be paid to evaluation of this system when it is introduced nationally. ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314310 5.6. Educational attainment Inconsistencies were apparent in overall results for the relationship between levels of education and job satisfaction. However, some findings contradicted the earlier research of Price and Mueller (1981) and Cavanagh (1992) by suggesting that higher educational attainment led to greater satisfaction at work and lowered intention to leave (Tzeng, 2002; Yin and Yang, 2002; Rambur et al., 2003). This contradiction may have occurred as a result of changes to nursing education that have developed over recent years in the countries where these studies were located (i.e. Australia, Singapore, Taiwan, US), most notably the introduction and expansion of degree-level qualifications. For example, Rambur et al’s (2003) US study, elucidates that baccalaureate education emphasises a greater under- standing of the context of health care delivery allowing for greater satisfaction. They suggest that it is this enhanced understanding that empowers nurses to influence the macro factors impacting on their work environment resulting in a reduction of dissatisfaction. If so, it could have relevance for the profession in the UK which, in an era of fast-paced change, may benefit from further expansion of degree education. However, the inconsistent results in both early and recent studies would suggest that educational attainment is a widely varying factor for which conclusions cannot yet be drawn and as such warrants further investigation. 5.7. Reflection on findings and model of nurse turnover behaviour The findings of this review were found to reflect the model of nurse turnover proposed by Irvine and Evans (1995). Although Irvine and Evans conducted their meta-analysis in a different employment climate to that which currently exists, the major findings of leadership/ supervisory relationship and stress remain inextricably linked to intent to leave. These factors comprise part of the work environment, which was found to have a stronger relationship with job satisfaction and intent to leave in both their meta-analysis and the findings of this paper. Changes over time have occurred regarding the effects on intent to leave for educational attainment and pay. This suggests that it is imperative that sources of satisfaction are reassessed in the light of changes over time. The empirical evidence shows that stress and issues concerning leadership consistently exert both direct and indirect effects on job satisfaction and intent to leave. Despite the improvement in techniques to analyse results from nurses’ job satisfaction studies, the reliance on quantitative methods for data collection has over- shadowed qualitative investigations (McNeese-Smith, 1999). Nevertheless, some qualitative inquiry has been used, often as a small component of a larger investiga- tion. This is exemplified in Cowin’s (2002) and Fletcher’s (2001) studies where additional written comments served to clarify issues raised in results from scale measure- ment. The information gained from qualitative inquiry could be utilised more widely to provide a relevant data source to complement quantitative methods. Not only would this serve to enhance contemporary and specific knowledge of nurses’ job satisfaction, but would also assist in providing the nursing profession with a methodology evolved from, but independent of, non- nursing areas. 5.8. Contextual factors Despite much research, definitive results of which variables are statistically and consistently related to nurses’ job satisfaction remain elusive (Tovey and Adams, 1999). The demonstration of variation in the factors influencing job satisfaction in different levels of the profession (e.g. of ward managers and staff nurses) and in differing environments (e.g. hospitals, commu- nities, geographical regions) reflects the difficulties faced when searching for authoritative results. Furthermore, there are discrepancies regarding sampling that involves mixing data from samples of nurses whose jobs have distinctive features such as paediatric, intensive care or mental health, as well as the inclusion of health care assistants. This lack of distinctiveness serves to invali- date generalisability of results to ward-based medical and surgical adult nurses. Similar problems exist with geographical locations that may have differing labour markets and population densities that may affect the opportunities for alternative employment or workloads respectively. Studies undertaken in differing types of hospitals offering different types of care (e.g. long-term or acute) may further produce results specific only to that environment, and therefore limit the relevance for the wider profession. Additionally, the generalisability of research based on theories and frameworks developed in one culture or nation and then undertaken in dissimilar countries has been questioned (Rosenzweig, 1994). However, in response to this, some international research on job satisfaction now offers comparisons across nations, contributing to empirical generalisation (Fang and Baba, 1993). Furthermore, the extent to which international recruitment has occurred over recent years and the paucity of research relating to immigrant nurses experiences of the British NHS necessitates the use of international data on job satisfaction to illuminate the views of nursing staff. 5.9. Implications for research Given the multitude of issues surrounding the study of job satisfaction components and their effects on intent ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 311 to leave, generalised conclusions should be drawn with caution. The use of scales developed for use on occupational groups other than nurses may no longer be appropriate, as the changing nature and specificity of work areas demand measurement approaches to be more exclusive. Research to identify particular stressors and effective ways of detecting stress early may be useful to address the impact that stress has on retention, and outcomes of the introduction of participative management for leadership dissatisfaction should be assessed. The literature selected for analysis contained no UK studies. This was unexpected and may indicate a discrepancy between the espoused importance of reten- tion issues and the UK’s research priorities and as such should be addressed and acted upon. Some work is, however, being undertaken such as the development of a tool for measuring job satisfaction in the contemporary UK nursing profession (Murrells et al., 2005), which will serve to accurately assess this exigent issue. 6. Conclusion This review set out to assess the components of job satisfaction most influential on intent to leave for nurses working with adults in ward environments. In addition to identifying these components, an exploration of how the sources of job satisfaction may have changed over time was undertaken. The results suggest that although many of the same factors are being assessed within scales, a more appropriate and relevant approach may be to assess each environment individually. Scales, whilst useful tools for comparability, need to evolve within the nursing profession for maximum effective- ness. If to be used as a means for implementing policy change and improving the working lives of nurses, it is essential that scales generate specific and accurate data. When the aim is improvement of job satisfaction for nurses, the conclusions made from this study point strongly toward the need for analysis at ward level. As differing results are found across work environments, more appropriate techniques such as qualitative inter- views or action research may be employed to gain detailed insight into which components are of impor- tance to particular workforces. This could be under- taken ward-wide, for policy change at NHS Trust level, as a nation-wide policy may not be apt given the discrepancies shown. Although consistent results were not always found, the issues of leadership and stress were highlighted as exerting particular influence on intent to leave. The implications of these issues remaining unresolved are detrimental for practice and the health service. References Aiken, L., Clarke, S., Sloane, D., Sochalski, J., Silber, J., 2002. Hospital nurse staffing and patient mortality, nurse burnout and job dissatisfaction. Journal of the American Medical Association 288 (16), 1987–1993. Bass, B., Avolio, B., 1995. Multifactor Leadership Question- naire (MLQ): Sampler Set Manual. Mind Garden, Rock- wood City. In: Larrabee, J., Janney, M., Ostrow, C., Withrow, M., Hobbs, G., Burant, C., 2003. Predicting registered nurse job satisfaction and intent to leave. Journal of Nursing Administration 33,(5) 271–283. Blegen, M., 1993. Nurses’ job satisfaction: a meta-analysis of related variables. Nursing Research 42 (1), 36–41. Blegen, M., Mueller, C., 1987. Nurses’ job satisfaction: a long- itudinal analysis. Research in Nursing and Health 10, 227–237. Bluedorn, A., 1978. A taxonomy of turnover. Academy of Management Journal 3, 647–651. Bluedorn, A., 1982. The theories of turnover: causes, effects and meaning. Research in the Sociology of Organisations 1, 75–128. Buchan, J., Seccombe, I., 2003. More Nurses, Working Differently?: A Review of the UK Nursing Labour Market 2002–2003. RCN, London. Campbell, J., Dunnette, M., Lawler, E., Weick, K., 1970. Management Behaviour, Performance and Effectiveness. McGraw-Hill, New York. Cavanagh, S., 1989. Nursing turnover: literature review and methodological critique. Journal of Advanced Nursing 14, 587–596. Cavanagh, S., 1992. Job satisfaction of nursing staff working in hospitals. Journal of Advanced Nursing 17, 704–711. Cavanagh, S., Coffin, D., 1992. Staff turnover among hospital nurses. Journal of Advanced Nursing 17, 1369–1376. Chiok Foong Loke, J., 2001. Leadership behaviours: effects on job satisfaction, productivity and organizational commit- ment. Journal of Nursing Management 9 (4), 191–204. Cowin, L., 2002. The effects of nurses’ job satisfaction on retention: an Australian perspective. Journal of Nursing Administration 32 (5), 283–291. Davidson, H., Folcarelli, P., Crawford, S., Dupart, L., Clifford, J., 1997. The effect of healthcare reforms on job satisfaction and voluntary turnover among hospital-based nurses. Medical Care 35 (6), 634–645. Demerouti, E., Bekker, A., Nachreiner, F., Schaufeli, W., 2000. A model of burnout and life satisfaction amongst nurses. Journal of Advanced Nursing 32 (2), 454–464. Department of Health, 1999. Improving Working Lives in the NHS. DH, London. Department of Health, 2000. The NHS Plan: A Plan for Investment, A Plan for Reform. DH, London. Department of Health, 2001. Investment and Reform for NHS Staff—Taking Forward the NHS Plan. DH, London. Department of Health, 2003. Agenda for Change: Proposed Agreement. DH, London. Fang, Y., 2001. Turnover propensity and its causes among Singapore nurses: an empirical study. International Journal of Human Resource Management 12 (5), 859–871. Fang, Y., Baba, V., 1993. Stress and turnover: a comparative study among nurses. International Journal of Comparative Sociology 34 (1–2), 24–38. ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314312 Finlayson, B., Dixon, J., Meadows, S., Blair, G., 2002. Mind the gap: the policy response to the NHS nursing shortage. British Medical Journal 325 (7363), 541–544. Fishbein, M., Ajzen, I., 1975. Belief, Attitude, Intention and Behaviour: An Introduction to Theory and Research. Addison-Wesley, Massachusetts. Fletcher, C., 2001. Hospital RN’s job satisfactions and dissatisfactions. Journal of Nursing Administration 31 (6), 324–331. Gauci-Borda, R., Norman, I., 1997. Factors influencing turn- over and absence of nurses: a research review. International Journal of Nursing Studies 34 (6), 385–394. Grimshaw, D., 2000. The problem with pay flexibility: changing practices in the UK health sector. International Journal of Human Resource Management 11 (5), 943–946. Hackman, J., Oldham, G., 1975. Development of the job diagnostic survey. Journal of Applied Psychology 60, 159–170. Hellman, C., 1997. Job satisfaction and intent to leave. Journal of Social Psychology 137, 677–689. Herzberg, F., 1966. Work and the Nature of Man. World Publishing, New York. Hinshaw, A., Atwood, J., 1985. Anticipated turnover among nursing staff study. Final report. National Institutes of Health, National Center for Nursing Research, Bethesda. Hinshaw, A., Smeltzer, C., Atwood, J., 1987. Innovative retention strategies for nursing staff. Journal of Nursing Administration 17 (6), 8–16. Irvine, D., Evans, M., 1995. Job satisfaction and turnover among nurses: integrating research findings across studies. Nursing Research 44 (4), 246–253. Kouzes, J., Posner, B., 2003. The Leadership Challenge, third ed. Jossey Bass, London. Larrabee, J., Janney, M., Ostrow, C., Withrow, M., Hobbs, G., Burant, C., 2003. Predicting registered nurse job satisfaction and intent to leave. Journal of Nursing Administration 33 (5), 271–283. Lu, K.-Y., Lin, P.-L., Wu, C.-M., Hsieh, Y.-L., Chang, Y.-Y., 2002. The relationships among turnover intentions, profes- sional commitment, and job satisfaction of hospital nurses. Journal of Professional Nursing 18 (4), 214–219. Maslow, A., 1970. Motivation and Personality, second ed. Harper and Row, New York. McKenna, E., 2000. Business Psychology and Organisational Behaviour: A Student’s Handbook, third ed. Psychology Press, Philadelphia. McNeese-Smith, D., 1993. Leadership behaviour and employee effectiveness. Nursing Management 24 (5), 38–39. McNeese-Smith, D., 1995. Job satisfaction, productivity and commitment. Journal of Nursing Administration 25, 17–26. McNeese-Smith, D., 1999. A content analysis of staff nurse descriptions of job satisfaction and dissatisfaction. Journal of Advanced Nursing 29 (6), 1332–1341. McVicar, A., 2003. Workplace stress in nursing: a literature review. Journal of Advanced Nursing 44 (6), 633–642. Meerabeau, E., Corby, S., Druker, J., White, G., 2004. Implementing local pay systems in nursing and midwifery. Journal of Advanced Nursing 47 (4), 368–376. Mobley, W., Horner, S., Hollingsworth, A., 1978. An evalua- tion of precursors of hospital employee turnover. Journal of Applied Psychology 63, 408–414. Morrison, R., Jones, L., Fuller, B., 1997. The relation between leadership style and empowerment on job satisfaction of nurses. Journal of Nursing Administration 27 (5), 27–34. Mueller, C., Price, J., 1990. Economic, psychological, and sociological determinants of voluntary turnover. Journal of Behavioural Economics 19, 321–335. Murrells, T., Clinton, M., Robinson, S., 2005. Job satisfaction in nursing: validation of a new instrument for the UK. Journal of Nursing Management 13, 296–311. Needleman, J., Buerhaus, P., 2003. Nurse staffing and patient safety: current knowledge and implications for action. International Journal of Quality Health Care 15, 275–277. Office of Manpower Economics, 2003. Workforce survey results for nursing staff, midwives & health visitors, 2003. www. ome.uk.com/downloads/NURSWF03.doc (last accessed 24/06/04). Prestholdt, P., Lane, I., Mathews, R., 1987. Nurse turnover as reasoned action: development of a process model. Journal of Applied Psychology 72, 221–227. Price, J., Mueller, C., 1981. Professional Turnover: The Case for Nurses. Iowa State University Press, Ames. Rambur, B., Palumbo, M., McIntosh, B., Mongeon, J., 2003. A statewide analysis of RN’s decision to leave their position. Nursing Outlook 51 (4), 182–188. Rosenzweig, P., 1994. When can management science research be generalised internationally? Management Science 40 (1), 28–39. Royal College of Nursing, 2002. Royal College of Nursing Congress 2002 report summary. Nursing Standard 16, 4–9. Saleh, S., Lee, R., Prien, E., 1965. Why nurses leave jobs. An analysis of female turnover. Personnel Administration 28, 25–28. Seccombe, I., Smith, G., 1997. Taking Part: Registered Nurses and the Labour Market in 1997. Institute for Employment Studies, Brighton. Shader, K., Broome, M., Broome, C., West, M., Nash, M., 2001. Factors influencing satisfaction and anticipated turn- over for nurses in an academic medical centre. Journal of Nursing Administration 31 (4), 210–216. Smith, P., Kendall, L., Hulin, C., 1969. Measurement of Satisfaction in Work and Retirement. Rand McNally, Chicago. Sorrentino, E., 1992. The effect of head nurse behaviours on nurse job satisfaction and performance. Hospital Health Services Administration 37, 103–113. Spector, P., 1997. Job Satisfaction: Application, Assessment, Causes and Consequences. Sage Publications, London. Stamps, P., Piedmonte, E., 1986. Nurses and Work Satisfac- tion: An Index for Work Satisfaction. Health Administra- tion Press, Michigan. Strachota, E., Normandin, P., O’Brien, N., Clary, M., Krukow, B., 2003. Reasons registered nurses leave or change employment status. Journal of Nursing Administration 33 (2), 111–117. Steele, R., Ovalle, N., 1984. A review and meta-analysis of research on the relationship between behavioural intentions and employee turnover. Journal of Applied Psychology 69, 673–686. Tovey, E., Adams, A., 1999. The changing nature of nurses’ job satisfaction: an exploration of sources of satisfaction ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314 313 in the 1990s. Journal of Advanced Nursing 30 (1), 150–158. Tzeng, H.-M., 2002. The influence of nurses’ working motiva- tion and job satisfaction on intention to quit: an empirical investigation in Taiwan. International Journal of Nursing Studies 39, 867–878. Upenieks, V., 2003. What constitutes effective leadership? Perceptions of magnet and non-magnet nurse leaders. Journal of Nursing Administration 33 (9), 456–467. Yin, J.-C.T., Yang, K.-P.A., 2002. Nursing turnover in Taiwan: a meta-analysis of related factors. International Journal of Nursing Studies 39, 573–581. ARTICLE IN PRESS B. Coomber, K. Louise Barriball / International Journal of Nursing Studies 44 (2007) 297–314314