A multilevel analysis of job satisfaction in Britain - WERS 2004

[Very Preliminary Draft]A multilevel analysis of workplace job satisfaction in Britain:evidence from a linked employer-employee dataGetinet Astatike Haile ∗Policy Studies Institute, London W1W 6UP; Tel: +44(0)2079117504;Email: haileg@psi.org.uk.January 2007AbstractJob satisfaction has been a fascinating theme of research in economics owing tothe link it is thought to have with such labour market behaviour of workers asturnover and absenteeism; and their performance and/or productivity. Whatdetermines job satisfaction? This paper undertakes a multilevel analysis of thedeterminants of job satisfaction in Britain using linked employer-employee datafrom the 2004 British Workplace Employment Relations Survey (WERS 2004).It departs from previous studies on several counts. First, it employs multilevelmodelling as opposed to the commonly used estimation technique in theliterature. Second, it investigates the determinants of up to eight different facetsof job satisfaction, in addition to overall job satisfaction, thereby being able toestablish what determines which aspects of job satisfaction. Third, unlike mostprevious studies it uses a whole host of correlates relating to employees and theirworkplaces. Not many studies have used linked employer-employee data to studyjob satisfaction, and the few studies that do have not exploited the nestedstructure in such data. By using linked employer-employee data and byexploiting the clustering therein this study is able to avoid the risk of misleading‘statistically significant’ effects. The study finds that 1) unmeasuredheterogeneity is important for the most part. 2) Controlling for such unmeasuredeffects, it finds that such factors as flexible work arrangement and better skillsmatch, among others, are crucial determinants of employees’ overall and otherfacets of job satisfaction.Key words: job satisfaction, multilevel analysis, linked employer-employee data.JEL classification: J28, I31∗ The author acknowledges very insightful discussions with Alex Bryson and Michael White.Helen Bewley has been very helpful with some aspects of the WERS 2004 data. Estimation of themultilevel models is performed using GLLAMM (http://www.gllamm.org) and the HighPerformance Cluster (HPC) facility at Lancaster University.

Acknowledgement"The author acknowledges the Department of Trade and Industry, the Economicand Social Research Council, the Advisory, Conciliation and Arbitration Serviceand the Policy Studies Institute as the originators of the 2004 WorkplaceEmployment Relations Survey data, and the Data Archive at the University ofEssex as the distributor of the data. The National Centre for Social Research wascommissioned to conduct the survey fieldwork on behalf of the sponsors. Noneof these organisations bears any responsibility for the author’s analysis andinterpretations of the data."2

1. IntroductionEconomists ventured into studying issues of subjective well-being and/or jobsatisfaction somewhat later than psychologists and sociologists. None the less,they have been fascinated by issues surrounding job satisfaction for it hasimportant economic consequences. In the economics circle job satisfaction hastraditionally been regarded as an important predictor of such labour marketbehaviours of workers as quit and absenteeism (e.g. Hammermesh 1997,Freeman 1978, Akerlof et al. 1988, Clark et al. 1998, Shields & Price 2002,Kristensen & Westergard-Nielsen 2004, Levy-Garbous et al. 2005). Both ineconomics and other areas of social research job satisfaction has also been linkedto individual well-being (Argyle 1989, Clark 1997, Sousa-Poza and Sousa-Poza2001, Kahneman & Krueger 2006), job performance and productivity (e.g.Iaffaldano & Muchinsky 1985), organisational performance (Ostroff 1992) andinnovation (Shipton, et al. 2006).There has been a growing interest in establishing the determinants of jobsatisfaction and many have argued as such (Hammermesh 2001). While mostprevious studies sought to relate job satisfaction to a host of worker and jobattributes (e.g. Clark 1996, Gazioglu and Tansel 2006), others have paid muchattention to some specific correlates of job satisfaction. These include, amongothers, job satisfaction and gender (Clark 1997, Bender et al. 2005), jobsatisfaction and age (Clark et al. 1996), job satisfaction and wage (Borjas 1979,Clark et al. 1998, Clark 1999, Lydon & Chevalier 2002), job satisfaction andrelative income (Clark and Oswald 1994, 1996), job satisfaction and unions(Borjas 1979, Berger et al. 1983, Bryson et al. 2004; 2006), job satisfaction andwork environment (Idson 1990), job satisfaction and work relations andmanagement (Gazioglu & Tansel 2004), job satisfaction and racial composition(Maume & Sebastian 2007).Though there exist such a long list of research into different aspects of jobsatisfaction and/or its consequences, there still exists a room for furthering ourunderstanding of the determinants of job satisfaction in particular. At least twoimportant reasons can be cited in this respect. First, almost all of the researchinto the determinants of job satisfaction fails to take into account unobservedindividual-level and/or workplace-level heterogeneity. 1 One important lesson thatresearch in labour economics, particularly where there is a linked/matchedemployer-employee data, underscores is the importance of such unobservedfactors in determining labour market outcomes (see, for example, Abowed et al.1999). 2 Job satisfaction as an outcome may not be too different in this regard,and one needs to account for unmeasured heterogeneity to address such issues as1 The few exceptions include Winkelmann & Winkelmann (1998), Clark et al. (2006), whoaccount for unmeasured effects using longitudinal data, and Bryson et al. (2004, 2006) who usethe 1998 sweep of WERS.2 Abowed et al. (1999) have used a matched longitudinal data. In WERS it is workplaces, and notemployees, that are followed in each of the surveys. Given this, the multilevel model offers thebest opportunity to account for employee-level and workplace-level heterogeneity in the contextof the cross-section (WERS 2004) data this study uses.3

worker sorting. Secondly, only few studies have explored the determinants ofvarious facets of job satisfaction, at least in the context of WERS, and nonelooked at eight different such facets. This, despite the importance attached toexploring the various facets of job satisfaction which has been stressed in somestudies (Rose 2001, Van Praag et al. 2002). Third, with few exceptions moststudies on job satisfaction have not used correlates of job satisfaction relatingboth to the employee and the workplace; and most of those that do fail to accountfor clustering in the data, which is likely to lead to misleading ‘statisticallysignificant’ effects.This study attempts to fill the gap in the existing literature in at least twoimportant ways. First, it uses linked employer-employee data and is able toaccount for a whole range of correlates of job satisfaction that relate to both theemployee and the employer. 3 Most studies in the job satisfaction literature havenot had the tantalising opportunity provided by the rich WERS 2004 data toanalyse the determinants of job and workplace satisfaction as this study does.Second and more importantly, this study uses a methodology that is mostappropriate to the data and the issue at hand; a methodology that exploits thenested structure of the WERS 2004 data, something that has not been usedpreviously. This allows extricating both observable and unobservable influenceson satisfaction at the employee and workplace level. That this paper explores upto eight different facets of job satisfaction alongside overall job satisfaction isalso another strength this study has. This enables determining whether jobsatisfaction is a single entity, as much of the literature makes it out to be, or not.The rest of the paper is organised as follows. In section 2, a brief review of theexisting literature will be made. In section 3, a description of the data andvariables used will be given. Section 4 sets out the empirical model used in theestimation. In section 5, results from the empirical exercise will be discussed.The final section concludes the paper.2. The literature on job satisfactionThe vast literature on satisfaction has attempted to establish the determinants ofjob satisfaction by modelling reported indicators of job satisfaction as a functionof a range of employee, employer and other job related correlates. The mostcommonly used such correlates include level of pay, hours of work, a whole hostof demographic and human capital characteristics, work environment, unionmembership. Though very few, there are also some studies that have used genderand racial segregation in a workplace as possible determinants of job satisfaction.This section attempts to make a brief review of the literature by way ofsummarising what it offers regarding the determinants of job satisfaction.With regards to demographic and human capital characteristics; the weight ofexisting evidence suggests a U-shaped relationship between age and jobsatisfaction (Clark, 1996, Clark et al. 1996, Sloane & Ward 2001) although thereis some evidence that satisfaction increases with age (e.g. Shields & Price 2002).The existing evidence also suggests that women are more satisfied with their job3 It is worth stating that some of these correlates are unique to the WERS 2004 data.4

than men (Clark, 1996, 1997; Clark & Oswald 1996, Blanchflower & Oswald1999; Groot & Brink 1999, Sloane & Williams 2000) but based on a corssnationalstudy Sousa-Poza & Sousa-Poza (2000) find this to be largely an Anglo-Saxon phenomenon. Higher level of education is generally associated with lowerlevel of job satisfaction (Clark 1996, Clark & Oswald 1996, Clark et al. 1996).Blanchflower & Oswald (2001) and Clark (1997) find some evidence attesting tohigher levels of job satisfaction for married individuals but Shield & Price (2002)find no such evidence. There is also some evidence relating to the effect of healthcondition on job satisfaction where Clark (1996) finds that health problems leadto lower level of job satisfaction.In terms of job related characteristics, income is probably the most widely usedcorrelate. Clark (1997) and Shields & Price (2002) find that income is animportant determinant of both overall job satisfaction and satisfaction with pay.Others, for example Clark and Oswald (1996), find weak correlation betweenabsolute income and job satisfaction while Belfield and Harris (2002) find noevidence that links job satisfaction with absolute income. However, there seemsa consensus on the link between ‘relative’ income and job satisfaction. 4 Anumber of studies including Clark and Oswald (1996), Levy-Garboua &Montmarquette (1997), Neumark & Postlewaite (1998), Sloane & Ward (2001),Hamermesh (2001), and Shield & Price (2002) find relative income as having aneffect on job satisfaction. Lydon & Chevalier (2002), on the other hand, questionthe validity of the notion of ‘relative’ income and estimate a satisfaction equationthat addresses issues of wage endogeneity, finding a strong effect of wages(current as well as future) on job satisfaction.As well as income and/or wage, hours of work has been found to impact jobsatisfaction. Clark (1996) finds hours to have a significant negative effect onoverall job satisfaction and an even stronger negative effect on satisfaction withpay. Union membership has, for the most part, been found to have a strongnegative effect on job satisfaction (Freeman 1978, Borjas 1979, Meng 1990,Miller 199). However, accounting for endogeneous selection as well asindividuals and firm level heterogeneity, Bryson et al. (2004; 2006) find nomarked difference in job satisfaction between unionized and non-unionizedworkers.With regards to workplace size the evidence is that larger establishment size isassociated with lower level of satisfaction (Idson 1990, Clark 1997, Gazioglu &Tansel 2006). Not much evidence exists regarding the relationship between jobsatisfaction and workplace segregations of different sorts. Bender et al. (2005)look into gender segregation and its effect on job satisfaction. They conclude thatprevious evidence suggesting an increase in the level of job satisfaction ofwomen as the share of women at the workplace increases was due to exclusion ofimportant determinants of jobs satisfaction such as flexibility between work andhome. Maume and Sebastian (2007) explore the effect of workgroup racialcomposition on the job satisfaction of white workers drawing on a broad 2002cross-section of workers in the US. 5 They find that an increase in the number of4 In Clark and Oswald (1996) ‘relative’ income is taken as the average income level of workerswith similar characteristics while Hammermesh (2001) regards ‘relative’ income as surprises thatmake an individual worker well off.5 The supposition here is that work is fundamentally a ‘social phenomenon’ and that it is at theworkplace, more than in other settings (such as neighbourhoods, schools, and churches) that5

minority co-workers affects the job satisfaction of whites adversely but that suchan effect vanishes once they control for the characteristics of jobs.3. The data and sampleThe data used in this study come from the 2004 British Workplace employmentrelations Survey, WERS 2004 (Department of Trade and Industry, 2005). It is themost recent survey in a series of five surveys of its kind. 6 The survey is regardedas one of the most authoritative sources of information on employment relationsin Great Britain. It offers linked employer-employee data drawn from apopulation of 700,000 workplaces (33 per cent of all workplaces in Britain) and22.5 million employees (89 per cent of all employees in employment) (Kersley etal. 2006). 7The sample of workplaces covered by the WERS 2004 survey is a randomsample of the 33 per cent of all workplaces in Britain stratified by workplace sizeand industry, with over-representation of large workplaces to make statisticallyreliable comparisons with their smaller counterparts. The probability of selectionof an employee in the survey is a product of the probability of the employee’sworkplace being selected and the probability of selection of the employee. Thisweight has been used in the empirical analysis in this study, to be able toextrapolate the findings to the population of workplaces in Britain with at leastfive employees (Kersley et al. 2006, Bryson et al. 2004).The WERS 2004 survey has covered a whole host of issues relating to bothemployers and employees, thereby allowing the inclusion of an array ofindividual and workplace level attributes into the analysis undertaken in thisstudy. The estimation sub-sample used in this study comprises of 20,692employees in some 1727 workplaces with complete information on all thevariables of interest that are summarised in Table A3 in the Appendix.The WERS 2004 survey asked each employee to rate, on a five-point scale from‘very satisfied’ to ‘very dissatisfied’, regarding how satisfied they were on eightaspects of their job including: (i) the sense of achievement they get from theirwork; (ii) the scope for using their own initiative; (iii) the amount of influencethey have over their job; (iv) the training they receive; (v) the amount of pay theyreceive; (vi) their job security; (vii) their involvement in decision making; and(viii) their work itself. An overall job satisfaction indicator has been generatedfrom these eight facets of job satisfaction in a way similar to that of Bryson et al.(2004). 8 For the purpose of the empirical analysis undertaken, the ordinalinter-group interaction exists between minorities and, in this case, whites; thereby impactingindividual job attitudes.6 Innovations in the most recent survey (WERS 2004) include, among others, the inclusion ofsmaller firms with between five and nine employees and the sampling of greater numbers of nonunionemployee representatives(Kersley et al. 2006).7 The survey population was all British workplaces but those in Agriculture, hunting & forestry,fishing, mining & quarrying, private households with employed persons, extra-territorialorganizations and bodies comprising of at least five employees (Kersley et al. 2006).8 Where a dummy variable equal to 1 if an employee was either ‘very satisfied’ or ‘satisfied’ and0 otherwise has been generated for each aspects of job satisfaction and eight such dummyvariables have been added to give an overall job satisfaction indicator.6

‘overall’ as well as each of the ‘facet’ satisfaction variables with five categorieshas been collapsed into a binary variable that assume a value one if an employeeis ‘very satisfied’ or ‘satisfied’ and zero otherwise. 9 Tables A1 and A2 in theAppendix report descriptive statistics and correlation matrix, respectively, of thesatisfaction variables.A summary of the explanatory variables used in the empirical modelling is alsogiven in Table A3 in the Appendix. These include: (i) employee-level correlatessuch as demographic and human capital characteristics, occupation, tenure, payas well correlates reflecting employees’ assessment of the degree of jobflexibility, job-level gender segregation and skill requirement of the job. (ii)workplace-level correlates such as establishment size, industry as well as otherspertaining to the level of gender-, age-, ethnic- and disability segregation at theworkplace.4. A framework of analysisJob satisfaction, or some aspect of it, is typically measured on an ordinal scale,requiring the use of ordered probit or logit as the preferred econometricmodelling strategy. This has led to the almost exclusive reliance on these modelsin the research into the determinants of job satisfaction (e.g. Clark & Oswald1994, Blanchflower 1996, Sousa-Poza & Sousa-Poza 2000, Gazioglu et al. 2006,Maume & Sebastian 2007). The consequence of this heavy reliance on thesemodels, or at least their simpler versions, has been the inability to account forunmeasured/unobserved heterogeneity, something that may be regarded as adrawback of the existing literature on job satisfaction. 10Accounting for such unmeasured heterogeneity is particularly important in thecontext of a linked employer-employee data used in this paper for at least twomain reasons. First, the variance of the estimated coefficients would beunderestimated, thereby leading to wrong inference. Secondly, parameterestimates would be inconsistent if the relationship between the outcome variableof interest and the explanatory variables is non-linear (Rodriguez and Goldman,1995, 2001; Skrondal and Rabe-Hesketh, 2004).As stated in the preceding section, each of the original ordinal satisfactionvariables used in the empirical analysis in this study has been collapsed into abinary variable that assumes a value one if ‘very satisfied’ or ‘satisfied’ and zerootherwise. 11 Thus, letting i index an employee, the underlying continuous latent9 The advantages (& drawback!) that stem from this are explained in the next section.10 There are few exceptions such as Winkelmann & Winkelmann (1998), Clark et al. (2006),Bryson et al. (2004, 2006)11 Although this approach raises the question of efficiency, not exploiting all information in theoriginal satisfaction variable(s), it makes the multilevel modelling much more tractable. That thisallows overcoming problems of estimator inconsistency stemming from the choice of breakingpoint in the ordered probability models (Crouchley 1995) and possible misclassification problemsarising from self-reported measures of job satisfaction that Hausman et al. (1998) refer to aresome of the problems the approach used in this study minimises.7

measure of overall or facet satisfaction of the employee can be represented by*which can be empirically modelled as S = α + β′x + ε , where x represents awhole array of correlates of job satisfaction.ii*iiiS is not observed in the data, ofcourse; instead we observe Siwhich assumes a value of 1 if S * i> 0 and 0otherwise.*SiIn a linked data of the sort used in this study, where employees are nested withinworkplaces, the most appropriate approach would be one that accounts for theclustering in the data which multilevel modelling does efficiently. Indexing aworkplace by j, we can re-specify Sito arrive at its multilevel equivalent Sijsothat;Sij= η + β x + β x + ε ,(1)0 j 1 ij 2 j ijwhere η0jare establishment specific intercepts, xijrepresents a employee levelcovariate having fixed effect β1,xjrepresents establishment specific covariateswith fixed effect β2and εijrepresent employee level residual terms. The η0 jaremodelled as,η = γ + ξ ,(2)0 j 00 0 jwhere the γ00is the mean intercept and ξ0 jrepresents the deviation of theestablishment specific intercept η from the mean. Letting θ ≡ Var(ε ) and0 jψ ≡ Var(ξ ), the two level random intercept multilevel model employed here0 jassumes that the clusters j are independent and satisfyε | ~ N(0, θ ),ijxijCov ( ε , ε') = 0, i ≠ i'iji jξ0|ij~ N(0, ψ ),jxCov(ξ 0, ) = 0.jε ijijSubstituting equation (2) into equation (1) gives the full model given bySij= γ + β x + β x + ξ + ε ,(3)00 1 ij 2 j 0 j ijso that the first three terms of equation (3) constitute the fixed effects of themodel while the last two terms form the random components of the same.5. Empirical results and discussionThe multilevel estimation results are reported in Table 1. As well as estimatingthe overall and each of the eight facets of job satisfaction for the full sample,estimations have been carried out separately for men and women for each8

measure of job satisfaction. 12 By way of simplifying the discussion of the largeset of empirical results reported in Table 1, this section is structured in such away that first, a discussion of the random effects will be made since establishingwhether these are important is one of the main aims of this study. Then willfollow the discussion of employee level fixed effects. Finally, findings relating toworkplace fixed effects will be discussed. In all cases, findings that hold acrossthe different measures of job satisfaction will be discussed first. This will then befollowed by a discussion of observed differences among the different measuresof job satisfaction which will in turn be followed by a discussion of possibledifferences between men and women in terms of their respective determinants ofjob satisfaction.Employee and workplace random effectsEstimates of the employee and workplace level random effects have been foundto be significantly different from zero in five of the nine models estimated,considering results for the full sample. The models where these effects are notsignificantly different from zero are: ‘overall satisfaction’, ‘satisfaction withsense of achievement’, ‘satisfaction with scope for own initiative’, and‘satisfaction with the work itself’. For men, the random effects become important/significantly different from zero/ in all but two of the models relating tosatisfaction with ‘sense of achievement’ and ‘satisfaction with pay’. For women,on the other hand, the random effects are important only for the models relatingto ‘satisfaction with influence on the job’, ‘satisfaction with trainingopportunity’, ‘satisfaction with job security’ and ‘satisfaction with involvementin decision making’.Employee fixed effectsWith regards to employee level fixed effects, the multilevel estimates relating toage, occupation, training opportunity, union membership, skill requirement of thejob vis-à-vis one’s own skill are some of the correlates found to have significanteffects on overall and the different facets of job satisfaction. In what follows,these effects are discussed with effects that run across the different modelsdiscussed first.Training opportunity: Employees that had some training opportunity up to 12months preceding the interview are significantly more likely to be satisfied attheir job, both in terms of overall satisfaction and most of its different facetsconsidered. The few exceptions where training is found not to have a significanteffect on satisfaction are: ‘satisfaction with pay’ for men and women combined;‘satisfaction with influence over the job’ and ‘satisfaction with pay’ for men, and‘satisfaction with scope for using own initiative’, ‘satisfaction with pay’ and‘satisfaction with the work itself’ for women.Union membership: Union membership is found to have a significant negativeeffect on overall job satisfaction and most of its facets modelled for the full12 This is done to be able to investigate whether there are differences between men and women interms of factors that determine overall and the different facets of job satisfaction, given thedifference in the labour market behaviour of men and women. The down side of this has ofcourse been the unusually large amount of estimation results from 27 different models reported inTable 1.9

sample. The exceptions to this for the full sample are ‘satisfaction with training’and ‘satisfaction with involvement in decision making’, where no significantunion effect is found. Relatively fewer significant union effects have been foundfor the models estimated separately for men and women, on the other hand. Formen, union has significant effect on ‘overall job satisfaction’ and four of theeight facets of job satisfaction that include ‘satisfaction with sense ofachievement’, ‘satisfaction with sense of influence on the job’, ‘satisfaction withjob security’ and ‘satisfaction with the work itself’. For women, on the otherhand, no significant effect of union membership has been found for ‘overall jobsatisfaction’ but union membership is found to have significant negative effecton five of the eight job facets namely, ‘satisfaction with the sense ofachievement’, ‘satisfaction with the scope for taking own initiative’, ‘satisfactionwith the level of influence on the job’, ‘satisfaction with pay’, and ‘satisfactionwith involvement in decision making’.Flexible working hours: Being able to work flexible hours renders by far themost number of significant effects in overall as well as the other facets of jobsatisfaction. This is the case for the full sample as well as the models estimatedseparately for men and women, where being able to work flexible hours is foundto have significant positive effects on job satisfaction. The only exception in thisregard is ‘satisfaction with the work itself’ for men, where being able to workflexible hours is not found to have a significant effect.Working from home: being able to work from home is another covariate found tohave a near universal significance. Being able to work from home is found tohave a significant positive effect for ‘overall job satisfaction’ and all its facetsexcept ‘satisfaction with training provisions’ and ‘satisfaction with job security’for the full sample. For men, on the other hand, being able to work from home isfound to have a significant positive effect for ‘overall job satisfaction’ and allfacets of job satisfaction but ‘satisfaction with training provisions’, ‘satisfactionwith pay’ and ‘satisfaction with job security’. For women, being able to workfrom home is found to have a significant positive effect on ‘overall jobsatisfaction’ and two of the eight facets of job satisfaction, namely, ‘satisfactionwith the level of influence on the job’ and ‘satisfaction with involvement indecision making’.Skills match/mismatch: Whether the skills required for one’s job match theemployee’s own skill is yet another covariate found to have universal effect onjob satisfaction, and the degree of universality appears to be stronger the higherthe skills mismatch. Thus, for the full sample, a higher mismatch between theskills required by jobs and employees’ skills is found to have a significantnegative effect on ‘overall job satisfaction’ and all other facets of jobsatisfaction. On the other hand, if the skills required by the job is only a ‘bithigher’ then the significant negative effect on job satisfaction is limited to‘overall job satisfaction’, ‘satisfaction with training’ and ‘satisfaction with pay’.In the models estimated separately for men and women, a similar pattern ofsignificance is found. For men, having skills that are much lower than the skillsrequired by the job is found to have a significant negative effect on ‘overall jobsatisfaction’ and all other facets of job satisfaction except ‘satisfaction withtraining opportunity at the workplace’ and ‘satisfaction with job security’. On the10

other hand, having skills that are a bit or much higher than the skills required bythe job is found to have a significant negative effect on ‘overall job satisfaction’,‘satisfaction with the sense of achievement’, ‘satisfaction with the scope forusing own initiative’. For women, working in jobs that require skills ‘muchhigher’ than own skills have a significant negative effect on ‘overall’ as well asall the other facets of job satisfaction. Being in jobs that require ‘a bit higher’level of skills than own skills, on the other hand, affect significantly negatively‘overall satisfaction’ and four of the eight facets of job satisfaction relating to‘satisfaction with training opportunity’, ‘satisfaction with pay’, ‘satisfaction withinvolvement in decision making’ and ‘satisfaction with the work itself’. Being ina job that requires ‘a bit or much’ lower level of skills than one’s own has asimilar negative and significant effect on job satisfaction for women. This is thecase for ‘overall’ job satisfaction and all facets of job satisfaction with theexception of ‘satisfaction with scope for taking own initiative’ and ‘satisfactionwith influence on the job’.Other employee level correlates: Other employee fixed effects are not found tohave significant effects across the satisfaction measures considered. However,they do provide some interesting reading. Two of the correlates that deserve amention in particular are employee’s occupation and level of pay. Accordingly,workers in ‘associate professions & technical occupations’, ‘administrative &secretarial occupations’, ‘personnel service occupations’ and ‘sales and customerservices occupations’ are significantly less likely to be satisfied with the‘influence they have on their job’ for the full sample, compared with ‘managersand senior officials’. Likewise, workers in all but ‘skilled trades’ occupations aresignificantly less likely to be satisfied with their ‘involvement in decisionmaking’ vis-à-vis their counterparts in ‘managers and senior officials’occupations for the full sample. These patterns of significance of occupation onjob satisfaction hold even strongly when we look at the estimates for men andwomen separately. In fact, in these models the observed significance holds for‘overall job satisfaction’ and few more facets of job satisfaction including thoserelating to ‘satisfaction with the sense of achievement’, ‘satisfaction with thescope for own initiative’, ‘satisfaction with pay’ and ‘satisfaction with the levelof security at the job’.In terms of the effect of the amount of gross weekly pay employees get on theirjob satisfaction, the estimation results reported in Table 1 indicate that comparedwith those that receive a weekly gross pay of between £141 and £220, those thatreceive a weekly pay of over £360 are significantly more likely to be satisfiedwith the amount of pay they receive for the full sample. The models estimatedseparately for men and women show a similar pattern in terms of the effect of theamount of weekly pay on the facet of job satisfaction relating to pay. Theadditional finding in these models is that those with a higher level of weekly payare also found to be significantly less likely to be satisfied with ‘the level oftraining’ they get at the workplace vis-à-vis their counterparts earning a weeklypay of between £141 and £220. 1313 However, there is a slightly odd finding that those that receive a weekly gross pay of at most£140 are also found to be significantly more likely to be satisfied with the amount of pay theyreceive compared with the reference group of workers that receive a weekly pay of between £14111

With regards to demographic characteristics of workers such as age, ethnicityand level of education; the estimated effects do not suggest any systematiceffects on job satisfaction. In terms of employee’s age, those that are 60 years orover are found to be significantly more likely to be satisfied with ‘overall’ aswell as three of the eight facets of job satisfaction relating to ‘satisfaction withthe sense of achievement’, ‘satisfaction with job security’ and ‘satisfaction withthe work itself’ compared with their counterparts aged between 40 and 49 yearswhen we consider the combined model of men and women. The modelsestimated separately for men and women show a similar effect of age onsatisfaction in that older workers (those 60 and over for men and those between50 and 59 years for women) are significantly more likely to be satisfied withsome facets of their jobs but not others. However, this age effect does not holdfor ‘overall’ job satisfaction when men and women are looked at separately.With regards to workers highest qualification, the multilevel estimates show thatthose with lower level qualifications, namely ‘No/other’ and ‘CSE/O-level’qualifications, are found to have a significant positive effect on satisfaction withthe ‘level of training they receive’ vis-à-vis their counterparts with a degree orhigher level qualification. This qualification effect on ‘satisfaction with training’holds both for the combined men and women model as well as the modelsestimated separately for men and women. For the men sample, lower level ofqualifications are also found to have a significant positive effect on satisfactionwith ‘scope for using own initiative’ compared with those with at least a degreelevel qualification.Workplace fixed effectsIndustry of employment: In terms of workplace fixed effects, the multilevelestimates that have a universal effect on job satisfaction are those that relate toindustry of employment. Accordingly, working in the education and healthindustries are found to have a significant positive effect on ‘overall’ and mostother facets of job satisfaction for the full sample, compared with theircounterparts working in the financial and business services industry. Though notas universal as the effects of being in the education and health industries, beingin the public & community services industry is also found to have significantpositive effect on four of the eight facets of jobs satisfaction. The effects of beingin the education and health industries observed for the full sample are repeatedfor men in the ‘overall’ and all other facets of job satisfaction except ‘satisfactionwith scope for own initiative’. For women, significant positive effects of being inthe education and health industries have been found for women with respect to‘overall’ and most other facets of job satisfaction. As well as these, being in thepublic & community services industry is also found to have a significant positiveeffect for most facets of job satisfaction for women.Other fixed effects: Though not as systematic as the industry effects, somesignificant effects relating to gender and ethnic composition of the workplacehave also been found.and £220 for the ‘overall’ and ‘pay’ facet of job satisfaction. This odd effect is repeated in theeffect that weekly pay has on ‘satisfaction with the amount of training’, for the women sample.12

6. ConclusionThis paper has investigated the determinants of job satisfaction using linkedemployer-employee data and a statistical technique that is best suited toanalysing such data. Unmeasured heterogeneity, employee-level and/oremployer-level, has largely been ignored in the job satisfaction literature whichthis paper attempted to address. As well as the unique data and methodology, thepaper has also studied the determinants of up to eight different facets of jobsatisfaction, alongside overall job satisfaction, by way of exploring whether jobsatisfaction is a single entity, as much of the literature assumes. It has alsoattempted to explore differences, if any, between men and women regarding thedeterminants of the different measure of job satisfaction.The study has some interesting findings that are of some relevance to the currentpolicy debate concerning the quality of working life in Britain. These include,among others: (i) the need for promoting flexible working (hours) arrangementand, where feasible, the opportunity to be able to work from home which arefound to impact the overall and other facets of job satisfaction significantlypositively. These are aspects of the working arrangement that can be fostered tobe able to promote well-being and all the positive consequences stemming fromit, something attainable given the current advance in IT and internet technology.(ii) the need for reducing the level of skills mismatch, say through on the jobtraining, is another issue that findings of this study point at. Employees in jobsthat require higher or lower levels of skills than the skills they own are found tobe significantly less likely to be satisfied in overall or the other faces of jobsatisfaction. Bridging such skills gaps is likely to foster overall and other facetsof job satisfaction and the benefits that accrue to the same. (iii) the findings inthis study, particularly those relating to ‘satisfaction with involvement indecision making’, point at occupational differences in involvement in decisionmaking. There seems a need for promoting workers’, both men and women,involvement in involvement in decision making, where there are no prohibitiveskill constraints, to raise this facet of job satisfaction and reaping the benefitthereof.13

Table 1 Multilevel estimates of overall & different facets of job satisfactionOverall satisfaction Sats. with sense of achievement Sats. with scope for using own initiativeAll Men Women All Men Women All Men WomenEmployee fixed effectsAge =60 0.547 0.548 0.547 1.120 1.146 1.025 0.453 0.636 0.316(0.263)* (0.359) (0.391) (0.293)** (0.350)** (0.483)* (0.263) (0.323)* (0.366)White 0.297 -0.082 0.743 -0.087 -0.284 0.148 0.050 -0.074 0.081(0.198) (0.299) (0.285)** (0.223) (0.308) (0.283) (0.185) (0.343) (0.193)No/other academic qualification -0.025 0.290 -0.323 0.021 0.017 -0.028 0.332 0.663 0.032(0.186) (0.237) (0.242) (0.231) (0.297) (0.286) (0.177) (0.232)** (0.255)CSE/O-level academic qual. 0.266 0.425 0.056 0.049 0.135 -0.124 0.327 0.620 0.072(0.178) (0.235) (0.208) (0.179) (0.229) (0.228) (0.154)* (0.211)** (0.200)A-level academic qualification -0.026 0.388 -0.370 -0.200 -0.021 -0.367 0.051 0.369 -0.214(0.166) (0.235) (0.239) (0.176) (0.207) (0.271) (0.157) (0.222) (0.206)Has dependent children 0.016 -0.241 0.315 -0.032 -0.109 0.078 0.184 -0.117 0.390(0.117) (0.176) (0.165) (0.127) (0.175) (0.184) (0.121) (0.175) (0.172)*Has other dependants -0.144 -0.244 -0.162 -0.191 -0.204 -0.257 0.036 -0.059 0.040(0.123) (0.207) (0.160) (0.133) (0.236) (0.175) (0.143) (0.235) (0.171)Has long-term illness or disabled -0.014 0.151 -0.151 0.005 -0.020 0.084 0.146 0.125 0.210(0.137) (0.184) (0.198) (0.146) (0.215) (0.196) (0.135) (0.192) (0.180)Professional occupations -0.435 -0.435 -0.317 -0.484 -0.846 0.151 -0.448 -0.654 0.077(0.266) (0.384) (0.317) (0.225)* (0.270)** (0.311) (0.241) (0.329)* (0.314)Associate prof. & technical occ. -0.763 -0.788 -0.709 -0.359 -0.888 0.152 -0.392 -0.515 -0.139(0.225)** (0.317)* (0.328)* (0.191) (0.265)** (0.263) (0.215) (0.284) (0.309)Administrative & secretarial occ. -0.873 -1.257 -0.534 -0.557 -0.888 -0.182 -0.465 -0.206 -0.260(0.230)** (0.347)** (0.304) (0.215)** (0.336)** (0.279) (0.198)* (0.292) (0.286)Skilled trades -0.093 -0.493 1.318 0.590 0.220 1.227 0.135 -0.176 1.331(0.306) (0.344) (0.676) (0.326) (0.369) (0.694) (0.322) (0.360) (0.665)*Personal service occupations -0.538 -0.503 -0.292 -0.076 -0.027 0.229 -0.294 -0.488 -0.026(0.262)* (0.407) (0.353) (0.268) (0.516) (0.309) (0.238) (0.362) (0.340)14

Sales and customer service occ. -0.843 -1.327 -0.393 -0.687 -1.278 -0.270 -0.348 -0.850 0.025(0.242)** (0.382)** (0.376) (0.215)** (0.322)** (0.287) (0.242) (0.379)* (0.337)Process, plant & machine oper. -0.723 -1.144 0.417 -0.631 -1.024 0.677 -0.952 -1.246 -0.269(0.280)** (0.333)** (0.537) (0.248)* (0.305)** (0.422) (0.288)** (0.349)** (0.441)Elementary occupations -0.895 -1.241 -0.462 -0.310 -0.490 -0.107 -0.356 -0.588 0.024(0.242)** (0.329)** (0.367) (0.234) (0.322) (0.312) (0.220) (0.289)* (0.338)On permanent contract -0.022 0.297 -0.308 0.334 0.739 -0.057 0.211 0.377 0.035(0.159) (0.246) (0.181) (0.161)* (0.252)** (0.201) (0.134) (0.238) (0.174)≥ 48 hrs at least once (12mnths) -0.160 -0.183 -0.066 0.024 -0.009 0.166 -0.045 -0.090 0.072(0.115) (0.166) (0.158) (0.128) (0.163) (0.173) (0.111) (0.175) (0.168)

(0.329) (0.311) (0.608) (0.390) (0.406) (0.616) (0.295)** (0.371)* (0.478)*Able to work flexible hrs 0.605 0.630 0.611 0.532 0.508 0.625 0.657 0.529 0.790(0.126)** (0.194)** (0.161)** (0.119)** (0.171)** (0.153)** (0.122)** (0.173)** (0.161)**DK if can work flexible hrs 0.293 0.414 0.211 0.203 0.154 0.258 0.232 0.210 0.341(0.125)* (0.195)* (0.161) (0.154) (0.238) (0.179) (0.137) (0.191) (0.173)*Can work from home 0.716 0.725 0.689 0.421 0.789 0.052 0.642 1.110 0.249(0.175)** (0.276)** (0.247)** (0.166)* (0.300)** (0.218) (0.174)** (0.269)** (0.214)DK can work from home -0.002 -0.181 0.105 0.094 0.326 -0.040 0.160 0.231 0.103(0.125) (0.217) (0.164) (0.143) (0.205) (0.189) (0.158) (0.230) (0.185)Job done only/mainly by men -0.321 0.011 -0.839 -0.167 0.146 -0.809 0.129 0.275 -0.248(0.141)* (0.172) (0.315)** (0.152) (0.190) (0.345)* (0.156) (0.190) (0.369)Job done only/mainly by wom -0.045 -0.729 -0.150 -0.083 -0.540 -0.191 -0.227 -0.781 -0.253(0.124) (0.287)* (0.135) (0.117) (0.337) (0.132) (0.118) (0.302)** (0.137)I am the only one doing the job -0.076 0.263 -0.252 0.149 0.498 0.111 0.238 0.183 0.355(0.197) (0.254) (0.250) (0.217) (0.292) (0.277) (0.223) (0.282) (0.318)Skill required much > own skill -1.161 -1.048 -1.207 -0.774 -0.734 -0.656 -0.606 -0.408 -0.705(0.125)** (0.176)** (0.169)** (0.140)** (0.192)** (0.184)** (0.132)** (0.193)* (0.167)**Skill required bit > own skill -0.279 -0.074 -0.411 -0.219 -0.079 -0.264 -0.126 0.037 -0.187(0.129)* (0.180) (0.166)* (0.119) (0.181) (0.147) (0.109) (0.160) (0.153)Skill required bit/much < own -1.271 -0.902 -1.527 -1.126 -0.872 -1.178 -0.546 -0.780 -0.378(0.230)** (0.353)* (0.324)** (0.218)** (0.330)** (0.278)** (0.242)* (0.351)* (0.308)Workplace fixed effects500 men working f-t 0.188 0.120 0.153 0.041 0.001 0.123 -0.320 -0.429 0.062(0.202) (0.247) (0.314) (0.204) (0.234) (0.249) (0.174) (0.258) (0.285)500 women working f-t -0.282 -0.471 -0.311 -0.126 0.054 -0.256 0.041 0.267 -0.295(0.239) (0.311) (0.310) (0.235) (0.303) (0.299) (0.228) (0.363) (0.308)100 men working p-t -0.121 0.307 -0.415 -0.396 -0.042 -0.494 -0.261 0.049 -0.38516

(0.269) (0.416) (0.298) (0.288) (0.457) (0.327) (0.247) (0.472) (0.296)100 women working p-t -0.044 -0.153 -0.093 0.277 -0.084 0.363 0.119 0.087 0.189(0.208) (0.360) (0.257) (0.220) (0.345) (0.280) (0.239) (0.365) (0.268)No worker aged 16-17 0.154 0.202 0.033 -0.320 -0.166 -0.411 -0.255 -0.132 -0.345(0.139) (0.194) (0.170) (0.151)* (0.198) (0.184)* (0.139) (0.174) (0.182)>5 workers aged 16-17 0.266 0.521 0.168 -0.487 0.128 -0.791 -0.144 -0.502 -0.068(0.161) (0.289) (0.186) (0.209)* (0.284) (0.242)** (0.155) (0.307) (0.202)No worker aged 18-21 -0.085 -0.367 0.075 -0.018 -0.154 -0.025 0.071 0.116 -0.016(0.125) (0.187)* (0.145) (0.116) (0.161) (0.156) (0.112) (0.181) (0.145)B/n 6-25 workers aged 18-21 -0.260 -0.502 -0.156 -0.221 -0.475 -0.181 -0.080 -0.104 -0.148(0.132)* (0.273) (0.240) (0.143) (0.244) (0.225) (0.155) (0.167) (0.223)>25 workers aged 18-21 0.004 0.066 -0.166 0.044 -0.031 -0.055 -0.021 0.280 -0.218(0.214) (0.296) (0.262) (0.210) (0.267) (0.252) (0.147) (0.231) (0.227)No worker aged over 50 -0.124 -0.303 -0.038 -0.100 -0.189 -0.043 -0.118 -0.300 0.020(0.156) (0.249) (0.194) (0.182) (0.227) (0.220) (0.143) (0.212) (0.177)B/n 6-25 workers over 50 -0.059 -0.218 0.070 0.142 0.058 0.216 -0.079 -0.292 0.051(0.123) (0.173) (0.165) (0.103) (0.159) (0.123) (0.107) (0.165) (0.141)>25 workers aged over 50 0.335 0.192 0.464 0.349 0.285 0.441 0.203 0.141 0.310(0.157)* (0.262) (0.206)* (0.163)* (0.266) (0.207)* (0.142) (0.256) (0.188)No ethnic minority worker 0.250 0.097 0.413 0.280 0.229 0.314 0.149 0.110 0.160(0.150) (0.231) (0.158)** (0.115)* (0.178) (0.135)* (0.109) (0.169) (0.152)Between 6-25 ethnic minority 0.174 0.135 0.179 0.307 0.494 0.021 0.295 0.543 0.106(0.153) (0.250) (0.138) (0.190) (0.287) (0.161) (0.169) (0.276)* (0.190)Over 25 eth minority workers 0.267 0.173 0.447 0.056 0.261 -0.141 -0.045 0.063 -0.192(0.157) (0.232) (0.222)* (0.160) (0.227) (0.250) (0.172) (0.232) (0.239)No disabled worker 0.193 0.150 0.269 0.185 0.239 0.210 0.239 0.376 0.239(0.126) (0.174) (0.154) (0.109) (0.176) (0.129) (0.127) (0.192)* (0.166)Over 5 disabled workers 0.283 0.215 0.255 0.478 0.487 0.473 0.254 0.331 0.192(0.162) (0.234) (0.201) (0.161)** (0.246)* (0.206)* (0.177) (0.245) (0.236)Manufacturing -0.154 0.030 -0.550 -0.195 -0.171 -0.278 0.012 0.140 -0.309(0.217) (0.245) (0.346) (0.234) (0.273) (0.324) (0.200) (0.252) (0.288)Construction 0.082 0.177 -0.061 -0.299 -0.057 -0.971 -0.391 -0.513 0.099(0.362) (0.405) (0.678) (0.313) (0.299) (0.700) (0.247) (0.374) (0.628)Trade, electricity, gas & water 0.309 0.533 0.235 0.364 0.389 0.462 -0.059 -0.154 0.020(0.153)* (0.246)* (0.247) (0.162)* (0.225) (0.241) (0.155) (0.224) (0.216)17

Hotels & rest, transp & comm. 0.243 0.343 0.232 -0.043 -0.026 0.059 -0.089 -0.094 -0.057(0.272) (0.376) (0.227) (0.205) (0.330) (0.242) (0.161) (0.261) (0.206)Public & community services 0.215 0.338 0.118 0.524 0.727 0.406 0.225 0.245 0.189(0.191) (0.233) (0.294) (0.183)** (0.253)** (0.245) (0.176) (0.281) (0.238)Education 0.720 1.242 0.633 1.405 1.503 1.397 1.093 1.528 0.979(0.210)** (0.611)* (0.233)** (0.246)** (0.478)** (0.270)** (0.230)** (0.407)** (0.243)**Health 0.677 0.652 0.676 0.839 0.594 0.899 0.521 0.204 0.660(0.201)** (0.360) (0.213)** (0.171)** (0.310) (0.198)** (0.171)** (0.363) (0.202)**Mgt views union favourably -0.132 0.041 -0.199 -0.281 0.001 -0.417 -0.417 -0.216 -0.500(0.144) (0.208) (0.165) (0.139)* (0.224) (0.170)* (0.134)** (0.213) (0.152)**Formal policy on equal opp. -0.132 -0.128 -0.099 -0.004 0.019 0.017 -0.131 -0.214 -0.045(0.146) (0.201) (0.168) (0.140) (0.178) (0.186) (0.112) (0.177) (0.146)Random effectsEmployee random effect -4.40e-11 1.14e-06 6.88e-11 1.82e-11 1.15e-11 1.07e-10 2.56e-11 -2.95e-06 1.66e-11(2.22e-07) (3.49e-07)** (2.30e-07) (1.45e-07) (3.94e-07) (1.50e-07) (8.11e-08) (4.22e-07)** (3.12e-07)Workplace random effect 1.08e-10 1.12e-06 9.57e-12 -1.07e-10 1.06e-10 -4.37e-11 -4.74e-11 -2.92e-06 8.80e-111.22e-07 (4.06e-07)** (2.87e-07) (2.61e-07) (5.43e-07) (2.01e-07) (9.76e-08) (4.23e-07)** (2.70e-07)Log-likelihood -2.08173 -.92821 -1.09117 -2.23667 -.9606 -1.2221 -2.2122 -.89188 -1.2681No. of workplaces 1727 1524 1600 1727 1524 1600 1727 1524 1600No. of employees 20692 9620 11072 20692 9620 11072 20692 9620 11072Standard errors in parentheses* significant at 5%; ** significant at 1%18

Table 1 (cont’d)Sats. with influence you have on the job Sats. with training you receive Sats. with amount of pay you getAll Men Women All Men Women All Men WomenEmployee fixed effectsAge =60 0.399 0.585 0.263 0.137 -0.237 0.490 0.279 0.341 0.353(0.246) (0.295)* (0.361) (0.214) (0.300) (0.315) (0.241) (0.350) (0.304)White -0.053 -0.743 0.627 -0.047 0.105 -0.137 0.252 0.082 0.454(0.216) (0.316)* (0.321) (0.215) (0.319) (0.213) (0.228) (0.306) (0.229)*No/other academic qualification 0.166 0.255 0.185 0.676 0.891 0.582 -0.154 -0.303 -0.039(0.173) (0.230) (0.241) (0.168)** (0.251)** (0.208)** (0.172) (0.245) (0.218)CSE/O-level academic qual. 0.123 0.206 0.102 0.417 0.436 0.455 0.082 -0.172 0.182(0.143) (0.206) (0.180) (0.143)** (0.220)* (0.175)** (0.134) (0.198) (0.172)A-level academic qualification 0.194 0.444 0.069 0.139 0.121 0.219 -0.208 -0.219 -0.294(0.140) (0.251) (0.178) (0.160) (0.249) (0.183) (0.153) (0.214) (0.195)Has dependent children 0.115 0.099 0.210 0.036 -0.082 0.243 -0.087 0.114 -0.190(0.100) (0.157) (0.142) (0.111) (0.148) (0.169) (0.113) (0.149) (0.150)Has other dependants -0.047 -0.060 -0.066 -0.288 -0.270 -0.364 -0.301 -0.417 -0.303(0.123) (0.185) (0.144) (0.126)* (0.194) (0.165)* (0.135)* (0.253) (0.153)*Has long-term illness or disabled -0.001 0.053 -0.061 -0.255 -0.298 -0.240 -0.292 -0.379 -0.183(0.148) (0.193) (0.191) (0.139) (0.209) (0.190) (0.130)* (0.203) (0.186)Professional occupations -0.697 -0.667 -0.599 0.088 -0.009 0.196 -0.647 -0.781 -0.622(0.199)** (0.274)* (0.308) (0.204) (0.264) (0.299) (0.187)** (0.259)** (0.281)*Associate prof. & technical occ. -0.770 -0.623 -0.886 -0.146 -0.022 -0.241 -0.235 -0.499 -0.047(0.187)** (0.260)* (0.270)** (0.170) (0.216) (0.226) (0.188) (0.243)* (0.261)Administrative & secretarial occ. -0.767 -0.606 -0.825 -0.225 -0.388 -0.084 0.028 -0.485 0.155(0.188)** (0.321) (0.261)** (0.178) (0.287) (0.249) (0.176) (0.264) (0.236)Skilled trades -0.190 -0.264 0.434 0.304 0.060 0.563 -0.693 -0.591 -0.476(0.250) (0.305) (0.546) (0.282) (0.295) (0.563) (0.271)* (0.285)* (0.551)Personal service occupations -0.968 -1.068 -0.917 -0.130 0.049 0.072 -0.449 -0.272 -0.399(0.243)** (0.334)** (0.320)** (0.219) (0.438) (0.284) (0.236) (0.315) (0.304)Sales and customer service occ. -0.516 -0.258 -0.656 0.123 -0.226 0.334 -0.015 -0.025 0.02719

(0.215)* (0.327) (0.303)* (0.208) (0.314) (0.280) (0.198) (0.292) (0.268)Process, plant & machine oper. -1.157 -1.151 -0.276 0.245 -0.018 0.674 0.011 -0.049 0.427(0.269)** (0.302)** (0.499) (0.226) (0.255) (0.439) (0.269) (0.315) (0.537)Elementary occupations -0.470 -0.442 -0.416 0.159 -0.134 0.439 -0.036 -0.054 0.138(0.210)* (0.292) (0.301) (0.222) (0.273) (0.333) (0.218) (0.312) (0.302)On permanent contract 0.296 0.372 0.187 0.124 0.253 0.015 -0.220 -0.001 -0.363(0.139)* (0.201) (0.182) (0.145) (0.218) (0.191) (0.151) (0.258) (0.190)≥ 48 hrs at least once (12mnths) -0.010 -0.114 0.114 -0.253 -0.255 -0.284 -0.349 -0.168 -0.508(0.110) (0.145) (0.147) (0.107)* (0.143) (0.148) (0.116)** (0.159) (0.167)**

Able to work flexible hrs 0.767 0.823 0.772 0.488 0.609 0.461 0.753 0.819 0.725(0.111)** (0.188)** (0.144)** (0.112)** (0.147)** (0.152)** (0.106)** (0.158)** (0.141)**DK if can work flexible hrs 0.248 0.414 0.215 0.197 0.238 0.191 0.578 0.805 0.390(0.122)* (0.192)* (0.156) (0.133) (0.185) (0.165) (0.135)** (0.222)** (0.156)*Can work from home 0.713 1.071 0.388 0.125 0.192 0.124 0.356 0.340 0.323(0.154)** (0.245)** (0.177)* (0.172) (0.200) (0.229) (0.159)* (0.216) (0.205)DK can work from home 0.200 0.196 0.185 -0.116 -0.013 -0.179 -0.018 -0.267 0.120(0.125) (0.216) (0.161) (0.115) (0.173) (0.153) (0.139) (0.187) (0.173)Job done only/mainly by men -0.021 0.166 -0.460 -0.275 -0.203 -0.345 0.152 0.268 0.451(0.162) (0.191) (0.304) (0.141) (0.155) (0.326) (0.158) (0.168) (0.349)Job done only/mainly by wom -0.180 -1.250 -0.125 -0.129 -0.743 -0.119 -0.152 -0.258 -0.298(0.108) (0.285)** (0.127) (0.113) (0.279)** (0.123) (0.118) (0.324) (0.139)*I am the only one doing the job 0.519 0.435 0.753 -0.331 0.078 -0.559 0.406 0.441 0.327(0.180)** (0.236) (0.252)** (0.193) (0.288) (0.238)* (0.167)* (0.231) (0.222)Skill required much > own skill -0.523 -0.606 -0.349 -0.442 -0.291 -0.571 -0.674 -0.562 -0.699(0.131)** (0.191)** (0.163)* (0.122)** (0.182) (0.156)** (0.127)** (0.197)** (0.160)**Skill required bit > own skill -0.010 0.078 -0.041 -0.270 -0.238 -0.318 -0.500 -0.004 -0.862(0.112) (0.165) (0.140) (0.100)** (0.150) (0.130)* (0.100)** (0.167) (0.151)**Skill required bit/much < own -0.500 -0.293 -0.510 -1.237 -0.842 -1.549 -0.468 0.303 -1.062(0.225)* (0.359) (0.302) (0.210)** (0.331)* (0.270)** (0.226)* (0.322) (0.265)**Workplace fixed effects500 men working f-t -0.167 -0.153 0.054 -0.114 -0.181 -0.130 -0.131 -0.209 0.016(0.195) (0.313) (0.252) (0.243) (0.246) (0.326) (0.176) (0.251) (0.317)500 women working f-t -0.166 -0.021 -0.498 -0.043 0.116 -0.177 -0.000 -0.556 0.149(0.197) (0.308) (0.253)* (0.238) (0.329) (0.295) (0.245) (0.308) (0.349)100 men working p-t -0.059 0.779 -0.514 -0.417 -0.117 -0.610 -0.552 -0.512 -0.740(0.211) (0.430) (0.229)* (0.241) (0.443) (0.262)* (0.349) (0.476) (0.455)21

100 women working p-t -0.028 -0.210 -0.062 -0.124 -0.772 0.156 -0.509 -0.224 -0.571(0.194) (0.309) (0.235) (0.211) (0.307)* (0.264) (0.205)* (0.313) (0.291)*No worker aged 16-17 -0.200 -0.250 -0.214 0.076 -0.142 0.182 -0.189 -0.457 -0.032(0.134) (0.179) (0.170) (0.151) (0.188) (0.182) (0.148) (0.199)* (0.175)>5 workers aged 16-17 -0.042 0.073 -0.034 -0.030 -0.554 0.186 0.332 -0.208 0.630(0.146) (0.257) (0.169) (0.202) (0.276)* (0.230) (0.293) (0.258) (0.355)No worker aged 18-21 -0.142 -0.422 0.008 -0.128 -0.296 0.020 -0.247 -0.195 -0.266(0.108) (0.167)* (0.131) (0.132) (0.165) (0.153) (0.128) (0.192) (0.143)B/n 6-25 workers aged 18-21 -0.208 -0.290 -0.231 -0.125 0.100 -0.176 -0.177 -0.015 -0.202(0.130) (0.204) (0.165) (0.282) (0.267) (0.319) (0.275) (0.222) (0.314)>25 workers aged 18-21 -0.163 -0.253 -0.322 -0.302 0.078 -0.525 -0.108 0.498 -0.516(0.148) (0.293) (0.182) (0.356) (0.358) (0.405) (0.239) (0.237)* (0.303)No worker aged over 50 -0.003 -0.079 -0.047 -0.018 -0.077 0.049 -0.075 -0.426 0.074(0.146) (0.228) (0.165) (0.175) (0.219) (0.208) (0.164) (0.218) (0.196)B/n 6-25 workers over 50 0.163 0.191 0.147 -0.217 -0.212 -0.228 -0.061 -0.232 0.030(0.102) (0.164) (0.129) (0.099)* (0.137) (0.128) (0.121) (0.179) (0.150)>25 workers aged over 50 0.327 0.597 0.142 -0.218 -0.102 -0.255 -0.056 -0.342 0.070(0.145)* (0.267)* (0.184) (0.156) (0.225) (0.208) (0.147) (0.224) (0.216)No ethnic minority worker 0.253 0.349 0.243 0.100 0.091 0.080 0.168 0.015 0.245(0.100)* (0.180) (0.124) (0.120) (0.171) (0.146) (0.115) (0.177) (0.137)Between 6-25 ethnic minority 0.215 0.278 0.132 0.018 0.094 -0.104 0.278 0.324 0.073(0.182) (0.259) (0.162) (0.174) (0.274) (0.149) (0.230) (0.312) (0.172)Over 25 eth minority workers -0.036 -0.153 0.116 -0.002 0.044 -0.000 0.070 -0.020 -0.050(0.162) (0.237) (0.209) (0.167) (0.226) (0.224) (0.202) (0.268) (0.268)No disabled worker 0.079 0.239 0.060 -0.142 -0.197 -0.033 -0.016 -0.091 0.073(0.116) (0.157) (0.138) (0.114) (0.168) (0.141) (0.131) (0.180) (0.165)Over 5 disabled workers 0.195 0.238 0.084 -0.060 -0.185 0.003 -0.186 -0.378 -0.044(0.173) (0.306) (0.191) (0.129) (0.184) (0.180) (0.169) (0.245) (0.251)Manufacturing 0.040 0.245 -0.382 0.188 0.479 -0.244 0.095 0.069 0.301(0.205) (0.239) (0.266) (0.237) (0.299) (0.298) (0.240) (0.282) (0.271)Construction 0.079 0.018 0.320 0.192 0.405 -0.012 0.317 0.239 1.054(0.298) (0.331) (0.605) (0.355) (0.355) (0.568) (0.235) (0.380) (0.534)*Trade, electricity, gas & water 0.248 0.135 0.401 0.116 0.188 0.120 -0.073 0.235 -0.191(0.159) (0.231) (0.208) (0.213) (0.258) (0.266) (0.195) (0.243) (0.232)Hotels & rest, transp & comm. -0.070 -0.139 -0.006 -0.051 0.004 -0.042 0.262 0.376 0.09722

(0.193) (0.332) (0.250) (0.292) (0.351) (0.326) (0.226) (0.332) (0.260)Public & community services 0.178 0.596 -0.013 0.371 0.403 0.318 0.557 0.676 0.607(0.151) (0.245)* (0.214) (0.183)* (0.226) (0.248) (0.169)** (0.255)** (0.217)**Education 0.869 1.396 0.719 0.388 0.709 0.213 1.354 1.649 1.283(0.169)** (0.425)** (0.225)** (0.177)* (0.271)** (0.227) (0.225)** (0.542)** (0.238)**Health 0.578 0.736 0.539 0.834 0.500 0.882 0.502 0.592 0.467(0.153)** (0.293)* (0.186)** (0.193)** (0.364) (0.224)** (0.188)** (0.261)* (0.215)*Mgt views union favourably -0.140 0.199 -0.275 0.033 0.187 -0.017 -0.059 0.048 -0.068(0.119) (0.177) (0.165) (0.118) (0.184) (0.148) (0.125) (0.177) (0.170)Formal policy on equal opp. -0.246 -0.178 -0.251 -0.012 -0.076 0.033 -0.368 -0.389 -0.337(0.119)* (0.186) (0.147) (0.141) (0.177) (0.174) (0.123)** (0.197)* (0.154)*Random effectsEmployee random effect -6.60e-06 -.0000123 -4.11e-06 -2.27e-06 -3.40e-06 -2.37e-06 -1.18e-06 -1.22e-07 -1.91e-08(1.49e-07)** (5.82e-07)** (1.97e-07)** (1.86e-07)** (5.23e-07)** (3.86e-07)** (2.54e-07)** (7.72e-07) (7.35e-07)Workplace random effect -6.41e-06 -.0000119 -4.00e-06 -2.20e-06 -3.29e-06 -2.30e-06 -1.17e-06 -1.21e-07 -2.35e-08(6.24e-07)** (1.04e-06)** (9.24e-07)** (6.57e-07)** (1.34e-06)** (9.95e-07)** (8.23e-07) (7.31e-07) (9.34e-07)Log-likelihood -2.5448 -1.0350 -1.4525 -2.6712 -1.1391 -1.4815 -2.5000 -1.0467 -1.3787No. of workplaces 1727 1524 1600 1727 1524 1600 1727 1524 1600No. of employees 20692 9620 11072 20692 9620 11072 20692 9620 11072Standard errors in parentheses* significant at 5%; ** significant at 1%23

Table 1 (cont’d)Sats. with your job security Sats. with your involvement in decision making Sats. with the work itselfAll Men Women All Men Women All Men WomenEmployee fixed effectsAge =60 0.485 0.457 0.436 0.267 0.181 0.264 0.851 1.054 0.566(0.240)* (0.332) (0.393) (0.259) (0.370) (0.324) (0.290)** (0.323)** (0.469)White -0.009 -0.192 0.145 0.292 0.458 0.307 -0.163 -0.575 0.249(0.237) (0.369) (0.267) (0.194) (0.285) (0.258) (0.200) (0.318) (0.234)No/other academic qualification 0.229 0.364 -0.001 0.061 -0.001 0.162 0.293 0.345 0.280(0.154) (0.214) (0.230) (0.153) (0.212) (0.210) (0.193) (0.257) (0.259)CSE/O-level academic qual. 0.200 0.126 0.140 0.219 0.315 0.191 0.085 0.318 -0.258(0.148) (0.214) (0.187) (0.134) (0.188) (0.177) (0.155) (0.210) (0.195)A-level academic qualification 0.126 0.189 0.078 0.139 0.388 0.005 -0.201 0.038 -0.428(0.145) (0.225) (0.202) (0.143) (0.243) (0.197) (0.166) (0.244) (0.213)*Has dependent children 0.112 0.047 0.187 0.180 0.046 0.268 0.039 -0.115 0.317(0.126) (0.172) (0.176) (0.100) (0.164) (0.147) (0.123) (0.170) (0.166)Has other dependants -0.305 -0.384 -0.348 -0.059 -0.248 0.010 -0.077 0.189 -0.348(0.127)* (0.225) (0.145)* (0.143) (0.280) (0.149) (0.145) (0.246) (0.163)*Has long-term illness or disabled -0.176 0.088 -0.331 -0.073 -0.171 -0.058 0.056 -0.248 0.438(0.152) (0.189) (0.197) (0.163) (0.193) (0.230) (0.134) (0.188) (0.210)*Professional occupations -0.393 -0.784 0.302 -0.870 -0.952 -0.759 -0.373 -0.473 -0.152(0.210) (0.264)** (0.301) (0.187)** (0.263)** (0.263)** (0.219) (0.255) (0.332)Associate prof. & technical occ. -0.459 -0.599 -0.309 -1.001 -0.998 -1.085 -0.168 -0.345 -0.018(0.176)** (0.242)* (0.273) (0.153)** (0.243)** (0.221)** (0.197) (0.266) (0.294)Administrative & secretarial occ. -0.373 -0.746 -0.211 -1.059 -0.754 -1.202 -0.327 -0.757 -0.078(0.178)* (0.264)** (0.250) (0.188)** (0.324)* (0.238)** (0.215) (0.336)* (0.299)Skilled trades -0.207 -0.266 0.122 -0.493 -0.559 -0.156 -0.037 -0.017 -0.923(0.295) (0.336) (0.601) (0.242)* (0.280)* (0.554) (0.327) (0.368) (0.503)Personal service occupations -0.273 -0.164 -0.016 -1.191 -1.369 -1.202 -0.076 -0.206 0.147(0.218) (0.336) (0.297) (0.227)** (0.431)** (0.307)** (0.273) (0.437) (0.335)Sales and customer service occ. -0.629 -0.725 -0.475 -0.958 -0.713 -1.175 -0.430 -0.816 -0.14824

(0.231)** (0.383) (0.306) (0.196)** (0.316)* (0.274)** (0.237) (0.325)* (0.315)Process, plant & machine oper. -0.347 -0.615 0.691 -1.015 -0.843 -1.253 -0.456 -0.477 -0.378(0.285) (0.305)* (0.553) (0.263)** (0.314)** (0.481)** (0.283) (0.333) (0.452)Elementary occupations -0.267 -0.611 0.415 -0.876 -0.745 -1.055 -0.183 -0.458 0.155(0.211) (0.287)* (0.301) (0.205)** (0.277)** (0.319)** (0.241) (0.313) (0.367)On permanent contract 1.055 0.877 1.301 -0.009 -0.034 -0.049 0.249 0.329 0.029(0.161)** (0.250)** (0.203)** (0.157) (0.233) (0.215) (0.147) (0.222) (0.206)≥ 48 hrs at least once (12mnths) -0.119 -0.023 -0.121 -0.139 -0.069 -0.238 -0.027 -0.074 0.189(0.110) (0.158) (0.162) (0.102) (0.143) (0.145) (0.124) (0.170) (0.164)

Able to work flexible hrs 0.716 1.067 0.440 0.533 0.395 0.663 0.326 0.252 0.483(0.124)** (0.177)** (0.165)** (0.108)** (0.161)* (0.150)** (0.117)** (0.184) (0.153)**DK if can work flexible hrs 0.411 0.593 0.236 0.159 0.099 0.257 0.410 0.283 0.590(0.143)** (0.213)** (0.174) (0.126) (0.192) (0.167) (0.140)** (0.178) (0.178)**Can work from home 0.224 0.202 0.145 0.636 1.019 0.341 0.625 0.996 0.195(0.159) (0.202) (0.213) (0.147)** (0.227)** (0.187) (0.179)** (0.254)** (0.234)DK can work from home -0.266 -0.459 -0.133 0.036 -0.010 0.053 -0.005 0.207 -0.182(0.139) (0.215)* (0.162) (0.125) (0.189) (0.172) (0.124) (0.210) (0.160)Job done only/mainly by men -0.122 0.173 -0.584 -0.380 -0.279 -0.970 -0.254 -0.073 -0.354(0.152) (0.155) (0.324) (0.158)* (0.176) (0.333)** (0.170) (0.198) (0.328)Job done only/mainly by wom 0.220 0.195 0.007 -0.211 -0.630 -0.183 -0.009 -1.035 -0.024(0.122) (0.307) (0.146) (0.111) (0.287)* (0.127) (0.130) (0.278)** (0.137)I am the only one doing the job 0.210 0.063 0.354 0.240 0.368 0.253 0.092 0.516 -0.107(0.204) (0.257) (0.267) (0.172) (0.231) (0.239) (0.230) (0.319) (0.287)Skill required much > own skill -0.439 -0.196 -0.573 -0.775 -0.765 -0.717 -0.860 -0.653 -1.046(0.113)** (0.167) (0.154)** (0.116)** (0.155)** (0.161)** (0.131)** (0.180)** (0.174)**Skill required bit > own skill -0.064 0.038 -0.094 -0.176 -0.024 -0.268 -0.171 0.127 -0.416(0.110) (0.167) (0.134) (0.113) (0.179) (0.134)* (0.127) (0.192) (0.163)*Skill required bit/much < own -0.861 -0.345 -1.250 -0.945 -0.551 -1.253 -0.985 -0.577 -1.363(0.237)** (0.390) (0.294)** (0.214)** (0.291) (0.281)** (0.238)** (0.357) (0.320)**Workplace fixed effects500 men working f-t -0.252 -0.258 -0.509 -0.079 -0.048 0.007 0.245 0.386 -0.073(0.185) (0.276) (0.259)* (0.192) (0.258) (0.345) (0.178) (0.261) (0.244)500 women working f-t -0.040 -0.313 0.085 -0.769 -1.205 -0.681 0.093 0.094 0.222(0.228) (0.395) (0.288) (0.215)** (0.347)** (0.317)* (0.210) (0.341) (0.249)100 men working p-t -0.301 -0.674 -0.109 -0.423 -0.319 -0.622 -0.226 -0.051 -0.413(0.257) (0.596) (0.264) (0.311) (0.532) (0.391) (0.265) (0.530) (0.286)26

100 women working p-t -0.361 -0.367 -0.314 0.094 0.351 -0.038 0.007 -0.311 0.107(0.196) (0.574) (0.266) (0.209) (0.358) (0.283) (0.174) (0.430) (0.204)No worker aged 16-17 -0.111 -0.290 -0.045 -0.175 -0.308 -0.112 -0.044 -0.128 -0.027(0.141) (0.202) (0.166) (0.137) (0.168) (0.187) (0.169) (0.256) (0.208)>5 workers aged 16-17 -0.113 0.382 -0.277 0.011 -0.274 0.116 -0.105 0.329 -0.320(0.193) (0.293) (0.194) (0.227) (0.319) (0.249) (0.191) (0.339) (0.222)No worker aged 18-21 -0.336 -0.407 -0.315 -0.047 -0.462 0.182 -0.209 -0.331 -0.182(0.125)** (0.171)* (0.150)* (0.111) (0.172)** (0.128) (0.118) (0.171) (0.164)B/n 6-25 workers aged 18-21 0.004 -0.362 0.212 -0.230 -0.276 -0.209 -0.043 -0.176 -0.018(0.128) (0.234) (0.191) (0.205) (0.160) (0.273) (0.182) (0.192) (0.262)>25 workers aged 18-21 0.392 0.114 0.492 -0.101 0.203 -0.327 -0.141 -0.456 -0.004(0.173)* (0.267) (0.212)* (0.260) (0.292) (0.308) (0.240) (0.484) (0.228)No worker aged over 50 -0.188 -0.529 0.014 -0.032 -0.224 0.081 -0.055 -0.160 0.048(0.157) (0.225)* (0.196) (0.160) (0.229) (0.184) (0.165) (0.248) (0.190)B/n 6-25 workers over 50 0.055 0.157 -0.044 -0.308 -0.330 -0.274 0.155 0.337 0.030(0.116) (0.162) (0.146) (0.115)** (0.172) (0.131)* (0.127) (0.170)* (0.181)>25 workers aged over 50 0.207 0.456 0.010 -0.271 -0.166 -0.258 0.453 0.848 0.153(0.172) (0.274) (0.196) (0.167) (0.304) (0.201) (0.158)** (0.262)** (0.207)No ethnic minority worker 0.212 0.226 0.117 0.259 0.334 0.209 0.397 0.352 0.414(0.140) (0.201) (0.171) (0.113)* (0.180) (0.125) (0.133)** (0.187) (0.159)**Between 6-25 ethnic minority 0.200 0.400 -0.099 0.073 0.260 -0.156 0.140 0.170 0.118(0.224) (0.293) (0.203) (0.177) (0.243) (0.197) (0.229) (0.332) (0.234)Over 25 eth minority workers -0.097 -0.189 -0.056 0.261 0.622 -0.014 0.068 0.025 0.064(0.167) (0.269) (0.185) (0.182) (0.256)* (0.248) (0.162) (0.231) (0.206)No disabled worker 0.080 -0.140 0.281 -0.136 0.103 -0.209 0.117 0.196 0.143(0.130) (0.179) (0.158) (0.147) (0.201) (0.164) (0.112) (0.202) (0.129)Over 5 disabled workers 0.266 0.254 0.329 0.037 0.135 0.049 0.219 0.240 0.222(0.237) (0.427) (0.219) (0.165) (0.301) (0.256) (0.176) (0.305) (0.180)Manufacturing -0.212 0.123 -0.517 0.201 -0.109 0.420 -0.208 -0.294 -0.074(0.192) (0.234) (0.387) (0.242) (0.291) (0.319) (0.161) (0.230) (0.290)Construction 0.204 0.261 0.712 0.927 0.617 1.283 -0.115 -0.176 -0.273(0.470) (0.483) (0.642) (0.354)** (0.450) (0.658) (0.345) (0.420) (0.585)Trade, electricity, gas & water 0.468 0.830 0.296 0.258 -0.082 0.510 0.328 0.448 0.329(0.161)** (0.220)** (0.245) (0.195) (0.263) (0.248)* (0.173) (0.245) (0.248)Hotels & rest, transp & comm. 0.128 0.453 -0.311 0.221 -0.145 0.542 -0.034 -0.324 0.25927

(0.223) (0.299) (0.293) (0.191) (0.315) (0.253)* (0.234) (0.311) (0.242)Public & community services 0.282 0.895 -0.141 0.152 -0.002 0.274 0.694 0.533 0.811(0.206) (0.284)** (0.256) (0.203) (0.320) (0.236) (0.180)** (0.271)* (0.246)**Education 0.832 1.793 0.163 0.851 1.197 0.773 1.194 1.765 1.168(0.212)** (0.621)** (0.222) (0.193)** (0.509)* (0.232)** (0.179)** (0.404)** (0.235)**Health 0.371 0.483 0.215 0.539 0.613 0.565 0.653 0.722 0.702(0.211) (0.339) (0.232) (0.197)** (0.336) (0.222)* (0.204)** (0.359)* (0.225)**Mgt views union favourably -0.223 -0.472 -0.042 -0.265 -0.060 -0.389 -0.449 -0.141 -0.609(0.134) (0.200)* (0.161) (0.132)* (0.197) (0.157)* (0.138)** (0.242) (0.158)**Formal policy on equal opp. -0.407 -0.425 -0.405 -0.138 -0.519 0.150 0.125 0.030 0.272(0.142)** (0.191)* (0.199)* (0.136) (0.201)** (0.160) (0.132) (0.188) (0.180)Random effectsEmployee random effect .00035 -9.46e-06 -.000015 -1.86e-06 -6.24e-06 -1.27e-06 5.96e-11 -.000015 -6.44e-11(.00002)** (3.40e-07)** (8.50e-07)** (1.69e-07)** (3.67e-07)** (4.58e-07)** (1.75e-07) (8.23e-07)** (2.01e-07)Workplace random effect .00034 -9.20e-06 -.000014 -1.80e-06 -6.08e-06 -1.23e-06 -4.49e-11 -.000014 9.23e-11(.00002)** (9.17e-07)** (1.01e-06)** (7.25e-07)** (3.53e-07)** (1.07e-06) (1.17e-07) (1.65e-06)** (2.84e-07)Log-likelihood -2.4315 -1.0721 -1.2848 -2.6053 -1.07800 -1.4706 -2.2011 -.9959 -1.1257No. of workplaces 1727 1524 1600 1727 1524 1600 1727 1524 1600No. of employees 20692 9620 1172 20692 9620 11072 20692 9620 11072Standard errors in parentheses* significant at 5%; ** significant at 1%28

ReferencesAbowd JM, Kramarz F and Margolis DN (1999) High wage workers and highwage firms, Econometrica, 67 (2), 251-333.Akrelof, GA, Rose, AK and Yellen JL (1988) Job Switching and job satisfactionin the US labour market, Brooking Papers on Economic Activity, 2, 495-582.Argyle M (1989) The Social Psychology of Work, 2 ndHarmondsworth.edition, Penguin,Belfield C and Harris R (2002) How well do theories of job matching explainvariation in job satisfaction across educational level? Evidence for UK graduates,Applied Economics, 34, 535-0548.Bender, KA, Donohue, SM and Heywood, JS (2005) Job Satisfaction and gendersegregation, Oxford Economic Papers, 57, 479-496.Berger, CJ, Olson CA and Boudreau, JW (1983) Effects of unions on jobsatisfaction: the role of work-related values and perceived rewards,Organizational Behaviour and Human Performance, 32, 289-324.Blanchflower D., and Oswald A. (1999) Well-being, insecurity and the decline ofAmerican job satisfaction, Mimeo, Dartmouth College, USA.Borjas G, (1979) Job satisfaction, wages and union, Human Resources, XIV, 21-40.Bryson, A. Cappellari, L and Lucifora, C (2004) Does Union Membership ReallyReduce Job satisfaction? British Journal of Industrial Relations.Bryson, A. Cappellari, L and Lucifora, C (2005) Why so unhappy? The Effectsof unionisation on job satisfaction, IZA discussion paper.Bryson, A. Cappellari, L and Lucifora, C (2006) Why so unhappy? The Effectsof unionisation on job satisfaction, Mimeo.Clark AE (1999) Are wages habit-forming? Evidence from micro data, Journalof Economic Behavior and Organization, 39, 179-200.Clark AE (2001) What really matters in a job? Hedonic measurement using quitdata, Labour Economics, 8, 223-242.Clark AE and Oswald AJ (1994) Unhappiness and unemployment, EconomicJournal, 104, 648-659.Clark AE, and Oswald A J (1996) Satisfaction and comparison income, Journalof Public Economics, 61, 359-381.Clark AE, and Oswald AJ (1996) Satisfaction and comparison income, Journalof Public Economics, 61, 359-381.Clark AE, Geogellis Y and Sanfey P (1997) Job satisfaction, wage changes andquits: Evidence from Germany, Research in Labour Economcis, 17, 95-121.Clark AE, Oswald A, and Warr P (1996) Is job satisfaction U-shaped in age?Journal of Occupational and Organizational Psychology, 69, 57-81.Clark, AE (1997) Job satisfaction and gender: Why are women so happy atwork? Labour Economics, 4, 341-372.29

Clark, AE, Georgellis, Y and Sanfey, P (1998) Job satisfaction, wage changes,and quits, Research in Labour Economics, 17, 95-121.Clark, AE, Kristensen, N and Westergard-Nielsen (2006) Jo Satisfaction and CoworkerWage: Status or Signalling? Working paper, Paris-Jourdan SciencesEconomiques (PSE).Crouchley, R (1995) A random-effects model for ordered categorical data,Journal of the American Statistical Association, 90, 489-498.Department of Trade and Industry (2005) Workplace Employment RelationsSurvey: Cross-Section, 2004 [computer file]. 1st ed. Colchester: The DataArchive [distributor], 21 December 2005. SN: 5294.Easterlin RA (2001) Income and happiness: Towards a unified theory, EconomicJournal, 111, July, 465-484.Freeman, RN (1978) Job satisfaction as an economic variable, AmericanEconomic Review, 68, 2, 135-141.Frey B and Stutzer A (2002) What can economist learn from happiness research?Journal of Economic Literature, 40, 402-435Galdeano AS (2001) Gender differences in job satisfaction and labour marketparticipation: UK evidence from propensity score, mimeo, EUI.Gazioglu S. and Tansel A (2004) Job satisfaction, work environment andrelations with managers in Britain, ERC working paper in economics, 03/04.Gazioglu S. and Tansel A (2006) Job satisfaction in Britain: individual and jobrelated factors, Applied Economics, 38, 1163-1171.Groot W and van den Brink H (1999) Job satisfaction and preference drift,Economic Letters, 63, 363-367.Haile, GA and Nguyen, AN (2002) Comparison income and job satisfaction:Evidence from GSOEP and NELS, Mimeo, Department of Economics, LancasterUniversity.Hamermesh D (1977), Economic aspects of job satisfaction, in Essays in LaborMarket Analysis, edited by Ashenfelter O and Oates W, Toronto: JohnWiley&Son.Hamermesh D (2001) ‘The changing distribution of job satisfaction’, HumanResources, 36 (1), 1-30.Hausman J, Abrevays J, and Scott-Morton, FM (1998) Misclassification of theDependent Variable in a Discrete-Response Setting, Journal of Econometrics,87, 239-269.Iaffaldano, MT and Muchinsky, PM (1985) Job Satisfaction and JobPerformance: A Meta-Analysis, Psychological Bulletin, 97 (2), 251-273.Idson, TL (1990), Establishment size, job satisfaction and the structure of work,Applied Economics, 22, 1007-10018.Kahneman, D and Krueger AB (2006) Developments in the Measurement ofSubjective Well-Being, Journal of Economic Perspective, 20 (1), 3-24.30

Kersley B., Alpin C., Forth J., Bryson A., Bewley H., Dix G. and Oxenbridge S.(2006), Inside the Workplace: Findings from the 2004 Workplace EmploymentRelations Survey, Routledge, London and New YorkKristensen, N and Westergard-Nielsen (2004) Does Low Job Satisfaction Lead toJob Mobility? IZA Discussion Paper No. 1026.Levy-Garboua L and Montmarquette C (1997), Reported job satisfaction: Whatdoes it mean? Cirano, Canada.Levy-Garboua L, Montmarquette C and Simonnet V (2001), Job satisfaction andquits: Theory and evidence from the German socioeconomic panel, Cirano,Canada.Lydon R and Chevalier A (2002) Estimates of the effect of wages on jobsatisfaction, Centre for Economic Performance, LSE.Maume DJ and Sebastian R (2007) Racial Composition of Workgroups and JobSatisfaction among Whites, The ANNALS of the American Academy of Politicaland Social Sciences, 609, 85-103.McBride M (2001) Relative-income effects on subjective well-being in the crosssection,Journal of Economic Behavior and Organization, 45, 251-278.Ostroff, C (1992) The Relationship Between Satisfaction, Attitudes, andPerformance: An Organization Level Analysis, Journal of Applied Psychology,77 (6), 963-974.Rodriguez, G and Goldman, N (2001) An Assessment of Estimation Proceduresfor Multilevel Models with Binary Responses, Journal of the Royal StatisticalSociety. Series A (Statistics in Society), 158 (1), 73-89.Rodriguez, G and Goldman, N (2001) Improved Estimation Procedures forMultilevel Models with Binary Response: A Case-Study, Journal of the RoyalStatistical Society. Series A (Statistics in Society), 164 (2), 339-355.Rose M (2005) Job Satisfaction in Britain: Coping with Complexity, BritishJournal of Industrial Relations, 43 (3), 455-467.Rose, M (2001) Disparate measures in the workplace … Quantifying overall jobsatisfaction, Paper presented at the 2001 British Household Panel SurveyResearch Conference, Colchester, UK.Shields M and Price S (2002) Racial harassment, job satisfaction and intentionsto quit: Evidence from British Nursing profession, Economica, 69, 295-362.Shipton HJ, West, MA, Parkes, CL and Dawson, JF (2006), When promotingpositive feelings pays: Aggregate job satisfaction, work design features, andinnovation in manufacturing organisations, European Journal of Work andOrganizational Psychology, 15 (4), 404-430.Skrondal, A. and Rabe-Hesketh (2004) Generalised Latent Variable Modelling:Multilevel, Longitudinal, and Structural Equation Models, Chapman &Hall/CRC.Sloane PJ and Williams H (2002) Job satisfaction, comparison earnings andgender, Labour, 14, 473-502.sSousa-Poza A and Sousa-Poza A (2000) Taking Another Look at theGender/Job-Satisfaction Paradox, KYKLOS, 53, 135-152.31

Sousa-Poza A and Sousa-Poza A (2000) Well-being at work: a cross-nationalanalysis of the levels and determinants of job satisfaction, Journal of Socioeconomics,29, 517-538.Usui, E (2006) Job Satisfaction and the Gender Composition of Jobs, Mimeo,Economics Department, Wayne State University.Van Praag BMS, Frijter P, and Ferrer-Carbonell A (2003) The anatomy ofsubjective well-being, Journal of Economic Behavior and Organization, 51, 29-49.Winkelmann L and Winkelmann R (1998) Why are the unemployed so unhappy?Evidence from panel data, Economica, 65, 1-15.32

AppendixTable A1 Descriptive statistics of the satisfaction/LHS variablesAll Men WomenLHS variables Mean St.dev Mean St.dev Mean St.devOverall job satisfaction (additive scale) 0.707 0.455 0.677 0.468 0.734 0.442Sense of achievement you get 0.698 0.459 0.670 0.470 0.723 0.448Scope for using own initiative 0.711 0.453 0.704 0.456 0.717 0.451Amount of influence you have over your job 0.566 0.496 0.562 0.496 0.569 0.495Training you receive 0.494 0.500 0.447 0.497 0.536 0.498Amount of pay you receive 0.353 0.478 0.337 0.473 0.366 0.483Job security 0.613 0.487 0.575 0.494 0.647 0.478Involvement in decision making you have 0.382 0.486 0.383 0.486 0.382 0.486The work itself 0.712 0.453 0.675 0.469 0.745 0.436No. of observations 20692 9620 11072Table A2 Correlation matrix of satisfaction variables (no. of obs 20692)1 2 3 4 5 6 7 8 91. Overall job satis 1.002. Sense of achievement 0.56 1.003. Scope for using own init. 0.55 0.51 1.004. Amount of influence you’ve 0.53 0.45 0.58 1.005. Training you receive 0.38 0.25 0.24 0.28 1.006. Amount of pay you receive 0.30 0.19 0.18 0.22 0.25 1.007. Job security 0.41 0.25 0.23 0.26 0.27 0.25 1.008. Involvement in decision mak’n 0.39 0.32 0.34 0.41 0.26 0.24 0.23 1.009. The work itself 0.54 0.58 0.44 0.41 0.26 0.22 0.29 0.29 1.0033

Table A3 Descriptive statistics of the RHS variablesAll Men WomenRHS variables Mean St.dev Mean St.dev Mean St.devEmployee characteristicsAge =60 0.043 0.203 0.055 0.229 0.032 0.177Female 0.535 0.499 -- -- -- --White 0.941 0.235 0.943 0.232 0.940 0.237No/other academic qualification 0.216 0.412 0.240 0.427 0.196 0.397CSE/O-level academic qualification 0.354 0.478 0.317 0.465 0.387 0.487A-level academic qualification 0.149 0.356 0.134 0.341 0.163 0.369Degree level or more academic qualification 0.280 0.449 0.310 0.462 0.255 0.436Has dependent children 0.394 0.489 0.414 0.493 0.375 0.484Has other dependants (to look after) 0.161 0.368 0.138 0.345 0.181 0.385Has long-term illness or disabled 0.118 0.323 0.130 0.336 0.108 0.310Managers and senior officials 0.114 0.317 0.156 0.363 0.077 0.266Professional occupations 0.123 0.328 0.134 0.340 0.114 0.317Associate professional and technical occupation 0.169 0.375 0.161 0.368 0.176 0.381Administrative and secretarial occupation 0.192 0.394 0.090 0.286 0.281 0.449Skilled trades 0.065 0.247 0.127 0.333 0.011 0.105Personal service occupations 0.087 0.282 0.033 0.180 0.133 0.340Sales and customer service occupations 0.069 0.254 0.039 0.195 0.095 0.293Process, plant and machine operatives 0.073 0.260 0.130 0.337 0.023 0.151Elementary occupations 0.108 0.311 0.128 0.334 0.091 0.287On permanent contract 0.919 0.272 0.927 0.260 0.912 0.283Has worked over 48 hrs at least once (last 12 months) 0.467 0.499 0.645 0.478 0.312 0.463

I am the only one doing the job 0.065 0.246 0.062 0.242 0.067 0.250Skill required for the job is much higher than own skill 0.209 0.407 0.243 0.429 0.180 0.384Skill required for the job is a bit higher than own skill 0.324 0.468 0.327 0.469 0.322 0.467Skill required for the job is about the same as own skill 0.418 0.493 0.375 0.484 0.455 0.498Skill required for the job is a bit/much lower than own 0.049 0.215 0.054 0.227 0.044 0.205Workplace characteristics500 men working full-time at the workplace 0.094 0.292 0.117 0.322 0.074 0.262500 women working full-time at the workplace 0.068 0.251 0.047 0.212 0.086 0.280100 men working part-time at the workplace 0.042 0.200 0.028 0.166 0.054 0.225100 women working part-time at the workplace 0.126 0.332 0.079 0.270 0.167 0.373No worker aged 16-17 at the workplace 0.708 0.455 0.722 0.448 0.696 0.460Up to 5 workers aged 16-17 at the workplace 0.190 0.392 0.193 0.395 0.187 0.390>5 workers aged 16-17 at the workplace 0.103 0.303 0.086 0.280 0.117 0.322No worker aged 18-21 at the workplace 0.302 0.459 0.279 0.449 0.323 0.467Up to 5 workers aged 18-21 at the workplace 0.332 0.471 0.335 0.472 0.329 0.470Between 6-25 workers aged 18-21 at the workplace 0.223 0.416 0.248 0.432 0.201 0.401>25 workers aged 18-21 at the workplace 0.143 0.350 0.137 0.344 0.147 0.354No worker aged over 50 0.078 0.268 0.072 0.258 0.083 0.276Up to 5 workers aged over 50 0.199 0.399 0.172 0.378 0.222 0.416Between 6-25 workers aged over 50 0.314 0.464 0.302 0.459 0.325 0.468>25 workers aged over 50 0.409 0.492 0.454 0.498 0.369 0.483No ethnic minority worker at the workplace 0.396 0.489 0.381 0.486 0.408 0.492Up to 5 ethnic minority workers at the workplace 0.302 0.459 0.317 0.465 0.290 0.454Between 6-25 ethnic minority workers at the workplace 0.160 0.367 0.165 0.371 0.157 0.364Over 25 ethnic minority workers at the workplace 0.141 0.348 0.137 0.344 0.145 0.352No disabled worker at the workplace 0.619 0.486 0.593 0.491 0.641 0.480Up to 5 disabled workers at the workplace 0.263 0.440 0.285 0.451 0.243 0.429Over 5 disabled workers at the workplace 0.119 0.323 0.122 0.327 0.116 0.320Manufacturing 0.147 0.354 0.233 0.423 0.073 0.259Construction 0.047 0.212 0.077 0.266 0.021 0.145Trade, electricity, gas and water 0.115 0.320 0.119 0.324 0.112 0.316Hotels and restaurants, transport & communication 0.088 0.283 0.117 0.321 0.062 0.242Financial and business services 0.181 0.385 0.181 0.385 0.180 0.385Public and community services 0.144 0.351 0.143 0.350 0.144 0.351Education 0.121 0.326 0.065 0.246 0.170 0.376Health 0.157 0.364 0.066 0.248 0.237 0.425Management views union membership favourably 0.387 0.487 0.354 0.478 0.415 0.493Workplace has equal opportunity/mgt diversity policy 0.899 0.302 0.882 0.323 0.913 0.281No. of observations 20692 9620 1107235