9
Int Arch Occup Environ Health (2012) 85:437–445 DOI 10.1007/s00420-011-0689-5 123 ORIGINAL ARTICLE The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement Sannie Thorsen · Reiner Rugulies · Katja Løngaard · Vilhelm Borg · Karsten Thielen · Jakob Bue Bjorner Received: 23 December 2010 / Accepted: 19 July 2011 / Published online: 10 August 2011 © Springer-Verlag 2011 Abstract Purpose The aim of this study was to examine the associ- ation between psychosocial factors (in particular ageism) at the workplace and older workers’ retirement plans, while taking health and workability of the employee into account. Methods In the fall and winter of 2008, self-report data on work environment, health, workability and retirement plans were collected in a representative national sample (n = 3,122) of Danish employees 50 years or older. Ordinal logistic regres- sion was used to analyse associations in a cross-sectional design. Predictor variables were standardized. Results In analyses adjusted for socio-demography, socio-economy, health, workability and work performance, 4 out of 6 examined psychosocial factors (ageism, lack of recognition, lack of development possibilities, lack of pre- dictability) were signiWcantly associated with plans of early retirement (OR: 1.10–1.13). StratiWed on gender, three psy- chosocial factors (ageism, lack of recognition, lack of development possibilities) remained signiWcant for men (OR: 1.15–1.25) and none for women. In particular was the association between retirement plans and ageism highly signiWcant in the male subgroup, but no association was found in the female subgroup. Conclusion Ageism, lack of recognition and lack of development possibilities are associated with older male workers’ retirement plans in our analyses. Workability has the strongest association with retirement plans for both genders. Keywords Age discrimination · Psychosocial work environment · Retirement plans · Work ability · Gender diVerences Introduction The demographic distribution in Western countries is changing due to the population’s longer lifespan and the falling birth rates, and this has led to a promotion of poli- cies to encourage the labour force participation of older workers in the European Union (Sigg and De-Luigi 2007). During the latest years, a reversal in the trend of early retirement has been achieved in several countries. In Denmark, however, changes have been small (Sigg and De-Luigi 2007). The reasons people retire from the labour market have been divided into 3 diVerent factors: (1) ‘push’ factors e.g. no appreciation or demand for older workers at the labour market, (2) ‘pull’ factors e.g. economical bene- Wts with early retirement, (3) ‘jump’ factors e.g. older per- sons’ desire to use the golden years in a new course of life (Jensen 2005; Blekesaune and Solem 2005). Several studies have found that economical beneWts of retiring pull people from the labour market and that reduced health and reduced workability push people out (Gruber and Wise 1999; Nielsen 1999; Sell et al. 2009; Tuomi et al. 1997). In addition, ageism has been blamed as a reason for early exits (Bennington 2001; Ilmarinen 1997). It has been argued that older people are pushed out of the labour market as part of a ‘institutionalised ageism’, where the possibilities and qualities of older workers are neither used nor recog- nized by managers, supervisors and employers (Ilmarinen S. Thorsen (&) · R. Rugulies · K. Løngaard · V. Borg · J. B. Bjorner The National Research Centre for the Working Environment, Lersø parkallé 105, 2100 Copenhagen, Denmark e-mail: [email protected] K. Thielen Department of Public Health, University of Copenhagen, Øster Farimagsgade 5, 1014 Copenhagen, Denmark

The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Embed Size (px)

Citation preview

Page 1: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Int Arch Occup Environ Health (2012) 85:437–445

DOI 10.1007/s00420-011-0689-5

ORIGINAL ARTICLE

The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Sannie Thorsen · Reiner Rugulies · Katja Løngaard · Vilhelm Borg · Karsten Thielen · Jakob Bue Bjorner

Received: 23 December 2010 / Accepted: 19 July 2011 / Published online: 10 August 2011© Springer-Verlag 2011

AbstractPurpose The aim of this study was to examine the associ-ation between psychosocial factors (in particular ageism) atthe workplace and older workers’ retirement plans, whiletaking health and workability of the employee into account.Methods In the fall and winter of 2008, self-report data onwork environment, health, workability and retirement planswere collected in a representative national sample (n = 3,122)of Danish employees 50 years or older. Ordinal logistic regres-sion was used to analyse associations in a cross-sectionaldesign. Predictor variables were standardized.Results In analyses adjusted for socio-demography,socio-economy, health, workability and work performance,4 out of 6 examined psychosocial factors (ageism, lack ofrecognition, lack of development possibilities, lack of pre-dictability) were signiWcantly associated with plans of earlyretirement (OR: 1.10–1.13). StratiWed on gender, three psy-chosocial factors (ageism, lack of recognition, lack ofdevelopment possibilities) remained signiWcant for men(OR: 1.15–1.25) and none for women. In particular was theassociation between retirement plans and ageism highlysigniWcant in the male subgroup, but no association wasfound in the female subgroup.Conclusion Ageism, lack of recognition and lack ofdevelopment possibilities are associated with older maleworkers’ retirement plans in our analyses. Workability has

the strongest association with retirement plans for bothgenders.

Keywords Age discrimination · Psychosocial work environment · Retirement plans · Work ability · Gender diVerences

Introduction

The demographic distribution in Western countries ischanging due to the population’s longer lifespan and thefalling birth rates, and this has led to a promotion of poli-cies to encourage the labour force participation of olderworkers in the European Union (Sigg and De-Luigi 2007).During the latest years, a reversal in the trend of earlyretirement has been achieved in several countries. InDenmark, however, changes have been small (Sigg andDe-Luigi 2007). The reasons people retire from the labourmarket have been divided into 3 diVerent factors: (1) ‘push’factors e.g. no appreciation or demand for older workers atthe labour market, (2) ‘pull’ factors e.g. economical bene-Wts with early retirement, (3) ‘jump’ factors e.g. older per-sons’ desire to use the golden years in a new course of life(Jensen 2005; Blekesaune and Solem 2005).

Several studies have found that economical beneWts ofretiring pull people from the labour market and that reducedhealth and reduced workability push people out (Gruberand Wise 1999; Nielsen 1999; Sell et al. 2009; Tuomi et al.1997). In addition, ageism has been blamed as a reason forearly exits (Bennington 2001; Ilmarinen 1997). It has beenargued that older people are pushed out of the labour market aspart of a ‘institutionalised ageism’, where the possibilitiesand qualities of older workers are neither used nor recog-nized by managers, supervisors and employers (Ilmarinen

S. Thorsen (&) · R. Rugulies · K. Løngaard · V. Borg · J. B. BjornerThe National Research Centre for the Working Environment, Lersø parkallé 105, 2100 Copenhagen, Denmarke-mail: [email protected]

K. ThielenDepartment of Public Health, University of Copenhagen, Øster Farimagsgade 5, 1014 Copenhagen, Denmark

123

Page 2: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

438 Int Arch Occup Environ Health (2012) 85:437–445

1997, 2006; Townsend 2006). Studies show that employershave less faith in older workers’ productivity and youngerworkers are preferred when hiring (Furunes and Mykletun2010; Riach and Rich 2006, 2007a, b; van Oorschot andJensen 2009). Older people are often seen in stereotypes(Loretto and White 2006), and even though if companieshave a ‘no age discrimination policy’, in practice evidenceof discrimination is found (Loretto and White 2006).

Few quantitative studies have examined older workers’own experience of ageism and to which degree they areleaving the work market due to perceived ageism. Onestudy among older Swedish workers (+55) has found thatperceived ageism was a signiWcant factor for early retire-ment among men, but not among women (Soidre 2005).

A poor psychosocial work environment may also bemore problematic for older workers than younger workers.It has been claimed that older workers have higher expecta-tions regarding the work climate than their younger coun-terparts and that they also feel higher obligations towardsthe workplace (Schalk 2004). Lack of respect and apprecia-tion of the older worker could be an important push factorfor the older workers’ exit from the labour market, even ifhis or hers workability is stable.

Some psychosocial work environment factors have beenlinked with retirement. Possibilities for development havebeen found to be associated with remaining at work or lateretirement (Lund and Borg 1999; Seitsamo 2005). Work-related conXicts have been shown to predict early retire-ment (Lund and Villadsen 2005). High job demands andlow job control as well as a mismatch between high jobdemands and low job control predicted retirement thoughtsor plans to retire early (Elovainio et al. 2005; Larsen 2008;Sutinen et al. 2005). Studies also exist that have failed toWnd associations between psychosocial factors and retire-ment. Among those are studies of emotional demands,management quality and relation to colleagues (Larsen2008; Lund and Villadsen 2005). It appears that physicaldemands are stronger linked with retirement than psychoso-cial job strains (Lund and Villadsen 2005).

In the above-mentioned studies, the analyses have beencontrolled for health of the employee but in no study wasthe analysis also controlled for workability. Workability isthe result of the interplay between the individual’sresources and the job demands (Ilmarinen 2007). If an olderemployee has a poor health and reduced workability, it mayboth aVect the employee’s experience of the psychosocialwork environment and the employee’s retirement plans. Itis also possible that a poor psychosocial work environmentaVects the health and workability of the employee, and thisleads to plans for early retirement. As Fig. 1 shows thesetwo pathways are not mutually exclusive.

In the present study, we aim to examine the associationbetween attitudes towards older workers (ageism) plus

other psychosocial work environment factors and plannedretirement. We examine the relative contribution fromhealth and workability compared to each psychosocial fac-tor by both reporting the eVects with and without adjust-ment for health and workability.

Methods

Population

Data were collected as part of the DAnish National workingEnvironment Survey (DANES) through late 2008 untilearly 2009. In the DANES, a subsample of employees50 years and older was asked questions regarding retire-ment plans and workability. This senior subsample wasextracted by Statistics Denmark as a simple random samplebased on the civil registration register (CPR-register) and aregister of all Danish employees. Due to regulatory require-ments, persons who previously had stated that they did notwish to participate in research surveys [approximately 8%for people above 50 years (Anonymous 2009)] wereexcluded from the random sample, resulting in a sample of4,477 persons. People in the sample received a question-naire by mail and were given the opportunity to answereither by internet or by mail. Non-responders received tworeminders and were Wnally contacted by phone.

Fig. 1 Model of the association between psycho-social work environ-ment and retirement plans

Retirement Plans

Psychosocial Work Environment

Ageism

Recognition

Possibilities for Development

Predictability

Quality of Leadership

Influence

Socio-economy related

Vocational Education

Occupational group

Social Class

Physical Work Strain

Age

Gender

Health related

Work Performance

Workability

Self-rated Health

Depressive Symptoms

123

Page 3: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Int Arch Occup Environ Health (2012) 85:437–445 439

The questionnaire was answered by 3,438 persons(77%). Of those, we excluded 306, because they answeredthat they were not working (n = 248), were self-employed(n = 45) or did not respond to the question regarding workstatus (n = 13). Demographic characteristics of the Wnal setof responders (N = 3,122) are shown in Table 1. Of theresponders, 75% used the postal questionnaire, 14% theinternet and 11% answered by telephone interview. Internet

and phone responders were re-contacted ap. 2 months aftertheir initial response to answer the question on plannedretirement, which was inadvertently left out of the originalquestionnaire.

The psychosocial work environment scales of the ques-tionnaire were tested for reliability in a test–retest setup. Inthe test–retest data, three data collection methods wereused, interviewed by phone, Wll out a paper questionnaire or

Table 1 Distribution of socio-demographic-, socio-economic- and health-related variables

n = 3,122 All Women Men

Socio-economy-related variables

Gender 53% 47%

Age in years (mean, SD) 56 (4) 56 (4) 56 (4)

Length of vocational education

<1 year 15% 15% 16%

¸1 and < 3 years 42% 40% 43%

¸3 and ·4 years 30% 36% 24%

> 4 years 12% 9% 17%

Social class

Higher professionals and managers 13% 7% 19%

Lower professionals and managers 20% 25% 15%

Higher clerical, services and sales workers 21% 29% 13%

Lower supervisors and technicians 2% 1% 4%

Lower clerical, services and sales workers 14% 21% 7%

Skilled workers 6% 1% 11%

Semi- and unskilled workers 24% 17% 32%

Physical work strain (levels 1–4) (mean, SD) 1.9 (0.9) 1.9 (0.9) 1.9 (1.0)

Occupational group

Human interaction jobs 34% 44% 23%

Information jobs 21% 21% 20%

Material/technical jobs 28% 15% 42%

Missing 17% 19% 15%

Health variables

SRH (levels 1–5) (mean, SD)

‘In general, would you say your health is: Excellent, very good, good, fair, poor’ 3.4 (0.8) 3.4 (0.8) 3.4 (0.8)

The Major Depression Inventory (levels 0–50) (mean, SD) 7.2 (7.1) 7.9 (7.4) 6.6 (6.7)

Work performance (levels 0–24) (mean, SD)

In comparison with others doing the same job as you, how would you rate your ability to: 13 (3.4) 13 (3.3) 14 (3.4)

(1) ‘manage a high work load?’

(2) ‘do your work well?’

(3) ‘work without making mistakes?’

(4) ‘make quick decisions?’

(5) ‘concentrate on work?’

(6) ‘How often was your work performance higher than others at your work place?’

Workability (levels 0–11) (mean, SD) 7.8 (2.2) 7.7 (2.2) 7.9 (2.3)

(1) ‘How do you rate your current workability with respect to the physical demands of your work?’

(2) ‘How do you rate your current workability with respect to the mental demands of your job?’

(3) ‘Is your workability reduced due to diseases, accident or detritions’

123

Page 4: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

440 Int Arch Occup Environ Health (2012) 85:437–445

answer a web version of the questionnaire. Time betweentest and retest was aimed at a 2-week period for all groups.In this study, we calculated reliability based on all test–retest data except those collected by phone. This resulted in1,085 test–retest respondents (response rate = 31%) where945 were employees.

Variables

The outcome variable of the study was planned retirementage: “When do you plan to retire?” with response catego-ries; (before age 60, at 60 or 61, between 62 and 65, at 65,at 66 or more) (see Table 2). Early retirement is consideredto be before the age of 65, the age from which old-age pen-sion is available in Denmark for this population.

The primary psychosocial work environment predictorwas perceived attitude to older workers (1) ageism: “Atyour workplace… Is there any space for elderly employ-ees?” The item comes originally from the ‘social inclusive-ness’ construct used in the Copenhagen PsychosocialQuestionnaire (COPSOQ) questionnaire (Pejtersen et al.2010).

Based on the COPSOQ questionnaire, we includeditems from Wve additional psychosocial constructs: ‘Recog-nition’ measured by a single item and ‘Possibilities fordevelopment’, ‘Predictability’, ‘Quality of leadership’ and‘InXuence’ that all are measured by two items each (seeTable 2). The reliabilities of the psychosocial constructswere estimated in a test–retest design. For the single items‘ageism’ and ‘recognition’, the weighted kappa was 0.46 and

Table 2 Distribution of the outcome ‘Retirement plans’ and the 6 psychosocial predictors

The variables are not standardized

All Women Men

Retirement: when do you plan to retire?

Before 60 3% 3% 3%

At 60 or 61 21% 28% 14%

Between 62 and 65 34% 33% 34%

At 65 14% 10% 19%

At 66 or more 9% 6% 12%

Undecided 19% 20% 18%

Ageism: is there any space for older employees?

To a very large extent 24% 24% 24%

To a large extent 49% 50% 48%

Some-what 22% 21% 23%

To a small extent 4% 4% 4%

To a very small extent 2% 2% 2%

Recognition: is your work recognized and appreciated by the management?

To a very large extent 16% 17% 15%

To a large extent 43% 44% 43%

Some-what 29% 28% 30%

To a small extent 9% 8% 10%

To a very small extent 3% 3% 3%

Lack of development possibilities (levels 1–5) (mean, SD)

(1) ‘Do you have the possibility of learning new things through your work?’ 2.4 (0.8) 2.4 (0.8) 2.4 (0.8)

(2) ‘Does your work require you to take initiative?’

Lack of inXuence (levels 1–5) (mean, SD)

(1) ‘Do you have a large degree of inXuence concerning your work’ 2.7 (1.0) 2.8 (1.0) 2.6 (1.0)

(2) ‘Can you inXuence the amount of work assigned to you?’

Poor quality of leadership (levels 1–5) (mean, SD)

(1) ‘To what extent would you say that your immediate superior - gives high priority to job satisfaction?’ 2.7 (0.9) 2.6 (0.9) 2.7 (0.9)

(2) ‘To what extent would you say that your immediate superior - is good at work planning?’

Lack of predictability (levels 1–5) (mean, SD)

(1) ‘Do you receive all the information you need in order to do your work well?’ 2.6 (0.9) 2.6 (0.8) 2.6 (0.9)

(2) ‘Are you informed well in advance for example concerning important decision, changes and plans for the future?’

123

Page 5: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Int Arch Occup Environ Health (2012) 85:437–445 441

0.58, indicating moderate agreement between test and retest(Landis and Koch 1977). For the remaining psychosocialconstructs, intraclass correlation (ICC) reliability between0.71 and 0.77 was found, thus above the general recom-mendation of at least 0.70 (Nunally and Bernstein 1994).

The health-related covariates were as follows: ‘Self-rated health (SRH)’ (Pejtersen et al. 2010), depressivesymptoms as measured by the ‘Major Depression Inven-tory’ (MDI) (Bech et al. 2001), self-assessment of ‘Work-ability’ and ‘Work performance’. The workability scale andthe work performance scale were developed through factoranalyses and diVerential item functioning analyses. Theworkability scale includes 3 items as follows: two questionsfrom the WorkAbility Index (WAI) (Ilmarinen 2007) andone item from the Danish Work Environment CohorteStudy (DWECS) (Burr et al. 2003) (see Table 1). The workperformance scale was assessed by 5 items designed for theDANES study and a 6th item derived from the Health andwork Performance Questionnaire (HPQ) (Kessler et al.2003) (see Table 1). The socio-economy covariates were asfollows: vocational education, social class (measured by theEuropean Socio-Economic ClassiWcation ESeC (2010)),physical strain at work and occupational group [(1) work-ing with customers, patients, pupils, etc, (2) working withinformation, (3) working with machines, material, animals,plants, etc, (4) missing job codes], age and sex (see Fig. 1).We used the Danish ISCO codes to classify the respondentsto one of these 3 occupational groups or the fourth groupwith missing job codes.

Statistical analysis

Some of the psychosocial work environment predictorswere substantial correlated (e.g. ‘job predictability’ and‘leadership quality’, Pearson correlation coeYcient = 0.68).If used in the same model, this may lead to multicollinear-ity and underestimation of the signiWcance of the predic-tors. The 6 psychosocial predictor variables were thereforeanalysed separately in the initial analyses. Each predictorwas standardized and entered into the model so high valuesof the predictor indicated poor work environment.

The categories of the outcome ‘planned retirement’ weretested for an ordinal relationship with multinomial logisticregression. Since these analyses supported ordinality, Wnalanalyses were performed using ordinal logistic regression.For each predictor, the analysis was performed in threesteps. In the Wrst step, each psychosocial predictor variablewas used as the only independent variable in a univariateanalysis. In step two, socio-economic and socio-demo-graphic variables and a data collection variable (post/web/phone) were included. In step three, the health- and work-ability-related covariates were also entered into the model.The signiWcant predictors from step 3 were Wnally entered

simultaneously into one model, thus controlling for eachother. Sensitivity analyses were performed stratiWed ongender and on occupational group.

Finally, the ordinal logistic regression analysis wasrepeated excluding respondents at and above 62 years ofage in order to address the possible bias of some responsecategories being irrelevant to those employees.

All analyses were performed with SAS 9.1.3.

Results

The outcome retirement plans

Twenty-four per cent of the respondents planned to retirebefore 62 years of age, 34% planned to retire between 62and 64 years of age, 23% wanted to retire at age 65 or laterand 19% were undecided on the subject. In gender-speciWcanalyses, 64% of the women and 51% of the men plannedto retire before 65 years of age (see Table 2).

The psychosocial predictors

The distribution of answers to the ‘ageism’ item “At yourworkplace…Is there any space for elderly employees?” wasthe following: 73% responded “true to a large” or “a verylarge extent”, 22% answered “somewhat” and 6% answered“to a small” or “a very small extent”. Distributions wereapproximately similar to men and women. Also, no majorgender diVerences were found for the other 5 psychosocialpredictors, although women felt a little more recognized atthe work place and experienced a little less inXuence (seeTable 2).

Main model ordinal regression (3 steps)

In the univariate ordinal logistic regression analyses (step1), all six psycho-social work environment factors were sig-niWcantly associated with planned retirement (see Table 3).In the gender-speciWc analyses, the associations were stron-ger for men than for women. In step 2, when all covariatesexcept SRH, depressive symptoms, workability and workperformance were included, all associations remained sta-tistically signiWcant in the combined dataset. In gender-spe-ciWc analyses, ‘Poor quality of leadership’ was not asigniWcant predictor in the female subgroup, but all otherpredictors remained signiWcant. In step 3 with all covariatesincluded, perceived ‘ageism’, ‘lack of development possi-bilities’, ‘lack of recognition’ and ‘lack of predictability’remained signiWcant in the combined dataset. ‘ageism’,‘lack of development possibilities’ and ‘lack of recogni-tion’ were signiWcant for men in the gender-speciWc analy-ses, and none were signiWcant for women.

123

Page 6: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

442 Int Arch Occup Environ Health (2012) 85:437–445

Psychosocial predictors in combined model

If the four signiWcant predictors were entered simulta-neously, only ‘lack of development possibilities’ remainedsigniWcant in the combined dataset, while only perceived‘ageism’ remained signiWcant for men (see Table 4).Among the covariates, ‘female gender’, ‘social class’, ‘poorworkability’ and ‘poor work performance’ were signiW-cantly associated with planned retirement in the combineddataset. ‘Social class’, ‘poor workability’ and ‘occupationalgroup’ were signiWcant in the gender-speciWc analysis formen and ‘physical work strain’ and ‘poor workability’ inthe gender-speciWc analysis for women. ‘Workability’ hadstrong association with planned retirement for both menand women. A further exploration of the results showedthat of the 3 items the workability scale is made of (see

Table 1), it is ‘physical workability’ that has the strongestassociation with planned retirement for both men andwomen, but all items in the scale were pointing in the samedirection (results not shown).

The eVect of occupation was explored in analyses strati-Wed on three occupational groups [(1) working with cus-tomers, patients, pupils, etc, (2) working with information,(3) working with machines, material, animals, plants, etc.].The 6 psychosocial predictors’ association with retirementplans was found in each of the three occupational groups.In the analyses were all covariates included. The resultshowed that within the human interaction jobs (N = 846respondents) only ‘lack of predictability’ was borderlinesigniWcant [OR (CI 95%) = 1.16 (1.00, 1.34)]. Within theinformation jobs (N = 492 respondents), only ‘lack of pos-sibilities for development’ was signiWcant [OR (CI

Table 3 Ordinal logistic regression of psychosocial work environments association with retirement plans

The psychosocial work environment predictors are standardized, and an increase responds to a worsening of the work environment. Only onepsycho-social predictor is included in each calculation of odds ratios

Bold values indicate associations with signiWcance levels below or at 0.05

1. step 2. step 3. step

No covariates Control for gender, age, vocational education, physical work strain, occupational group, ESeC, data collection mode

Control for all #2 covariates plus SRH, depressive symptoms, work performance and workability

OR (CI 95%) P OR (CI 95%) P OR (CI 95%) P

All data N = 2,358

Ageism 1.25 (1.17, 1.35) <0.0001 1.21 (1.12, 1.31) <0.0001 1.12 (1.03, 1.21) 0.01

Lack of recognition 1.26 (1.16, 1.36) <0.0001 1.21 (1.12, 1.31) <0.0001 1.09 (1.00, 1.19) 0.04

Lack of possibilities for development 1.25 (1.16, 1.34) <0.0001 1.20 (1.10, 1.31) <0.0001 1.13 (1.03, 1.23) 0.01

Low inXuence 1.33 (1.23, 1.43) <0.0001 1.15 (1.06, 1.24) 0.0008 1.05 (0.97, 1.15) 0.23

Poor leadership 1.17 (1.08, 1.26) <0.0001 1.14 (1.05, 1.23) 0.002 1.04 (0.96, 1.13) 0.37

Lack of predictability 1.25 (1.16, 1.35) <0.0001 1.21 (1.11, 1.31) <0.0001 1.10 (1.01, 1.20) 0.03

Men N = 1,121

Ageism 1.38 (1.24, 1.54) <0.0001 1.34 (1.19, 1.50) <0.0001 1.25 (1.10, 1.41) 0.0005

Lack of recognition 1.38 (1.23, 1.55) <0.0001 1.30 (1.15, 1.46) <0.0001 1.15 (1.01, 1.31) 0.03

Lack of possibilities for development 1.33 (1.19, 1.47) <0.0001 1.24 (1.10, 1.40) 0.0006 1.15 (1.02, 1.31) 0.03

Low inXuence 1.31 (1.18, 1.45) <0.0001 1.15 (1.02, 1.29) 0.02 1.07 (0.95, 1.21) 0.26

Poor leadership 1.26 (1.13, 1.40) <0.0001 1.19 (1.06, 1.34) 0.003 1.07 (0.95, 1.22) 0.28

Lack of predictability 1.31 (1.18, 1.46) <0.0001 1.23 (1.09, 1.37) 0.0005 1.11 (0.98, 1.26) 0.09

Women N = 1,237

Ageism 1.17 (1.05, 1.29) 0.003 1.12 (1.00, 1.25) 0.05 1.02 (0.91, 1.15) 0.73

Lack of recognition 1.21 (1.08, 1.34) 0.0006 1.15 (1.03, 1.28) 0.01 1.05 (0.94, 1.18) 0.36

Lack of possibilities for development 1.24 (1.11, 1.38) <0.0001 1.17 (1.03, 1.32) 0.01 1.09 (0.96, 1.24) 0.17

Low inXuence 1.24 (1.11, 1.38) <0.0001 1.15 (1.03, 1.29) 0.01 1.04 (0.92, 1.17) 0.52

Poor leadership 1.14 (1.03, 1.27) 0.01 1.08 (0.97, 1.21) 0.15 1.00 (0.89, 1.12) 0.98

Lack of predictability 1.21 (1.09, 1.35) 0.0005 1.19 (1.06, 1.33) 0.003 1.08 (0.96, 1.22) 0.21

123

Page 7: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Int Arch Occup Environ Health (2012) 85:437–445 443

95%) = 1.37 (1.09, 1.72)] and within material/technicaljobs (N = 663 respondents) only ageism was signiWcant[OR (CI 95%) = 1.22 (1.05, 1.42)].

Robustness

Older age predicted late retirement in all analyses (personsstill employed at 63 cannot plan to retire at 62). To testrobustness of the results, we repeated the data analysis ofTable 4 for people below the age of 62. The result wascomparable to the result from the analysis with all age clas-ses included. The main diVerences were that in the com-bined dataset the odds ratio for ‘lack of possibilities fordevelopment’ increased from 1.11 to 1.13 and in the analy-sis for the male subgroup the odds ratio for ageism decreasedfrom 1.20 to 1.16. No predictors changed from being sig-niWcant to non-signiWcant or vice versa.

Discussion

In this representative study of older Danish employees, 58%planned to retire before the age of 65, the age the Danish socialsecurity system oVers old-age pension. Early retirement is pos-sible through occupational early retirement pension (efterløn)from the age of 60 for this population (Anonymous 2010).

The hypothesis of the study (i.e. psychosocial work envi-ronment factors are associated with retirement plans) wassupported in the combined dataset. As we had expected,associations were attenuated when we adjusted for health,work performance and workability. But even after theadjustments, four out of six psychosocial factors (i.e. ‘age-ism’, ‘recognition’, ‘possibilities for development’ and‘predictability’) remained signiWcant at the 0.05 level.

In a gender-stratiWed analysis, with all covariatesincluded, the three factors ‘ageism’, ‘recognition’ and

Table 4 Ordinal logistic regression of psychosocial work environments association with retirement plans

Four psychosocial predictors and all covariates are included in the three (all data, men, women) analyses

Bold values indicate associations with signiWcance levels below or at 0.05

* Variables which are standardized

All data Men Women

OR (CI 95%) P OR (CI 95%) P OR (CI 95%) P

Predictors

Ageism* 1.08 (0.99, 1.18) 0.08 1.20 (1.05, 1.37) 0.01 1.00 (0.89, 1.13) 0.96

Lack of possibilities for development* 1.11 (1.01, 1.22) 0.04 1.11 (0.97, 1.27) 0.13 1.09 (0.95, 1.25) 0.21

Lack of recognition* 1.02 (0.92, 1.14) 0.65 1.08 (0.92, 1.26) 0.36 1.01 (0.87, 1.16) 0.94

Lack of predictability* 1.04 (0.93, 1.16) 0.48 1.01 (0.86, 1.17) 0.93 1.05 (0.90, 1.23) 0.51

Covariates

Gender males versus females 0.47 (0.39, 0.56) <0.0001

Low workability* 1.47 (1.31, 1.64) <0.0001 1.46 (1.24, 1.72) <0.0001 1.49 (1.28, 1.75) <0.0001

Low work performance* 1.11 (1.02, 1.20) 0.02 1.13 (1.00, 1.28) 0.05 1.07 (0.94, 1.20) 0.30

Depressive symptoms* 1.00 (0.91, 1.10) 0.94 0.92 (0.79, 1.08) 0.30 1.03 (0.91, 1.16) 0.67

Poor SRH* 1.08 (0.98, 1.20) 0.13 1.11 (0.95, 1.29) 0.19 1.06 (0.92, 1.21) 0.45

Vocational education (8 levels) (higher level = higher education)

0.96 (0.91, 1.00) 0.07 0.95 (0.88, 1.01) 0.11 0.96 (0.90, 1.03) 0.28

Physical work strain* 1.08 (0.98, 1.20) 0.11 1.00 (0.86, 1.15) 0.96 1.18 (1.02, 1.36) 0.03

ESeC (9 levels) (higher level = lower class) 1.05 (1.01, 1.09) 0.03 1.11 (1.04, 1.18) 0.002 1.02 (0.96, 1.08) 0.61

Type of data collection

Phone versus post 1.34 (1.00, 1.79) 0.05 1.30 (0.83, 2.02) 0.25 1.41 (0.95, 2.10) 0.09

Web versus post 0.84 (0.66, 1.07) 0.16 0.80 (0.58, 1.11) 0.18 0.89 (0.61, 1.31) 0.57

Ageclass

(64 and up) versus (50–52) 0.01 (0.01, 0.02) <0.0001 0.01 (0.01, 0.03) <0.0001 0.01 (0.00, 0.02) <0.0001

(60–63) versus (50–52) 0.24 (0.18, 0.31) <0.0001 0.23 (0.16, 0.34) <0.0001 0.22 (0.15, 0.33) <0.0001

(56–59) versus (50–52) 0.88 (0.71, 1.10) 0.26 0.66 (0.47, 0.93) 0.02 1.13 (0.84, 1.53) 0.42

(53–55) versus (50–52) 1.14 (0.90, 1.44) 0.27 0.90 (0.63, 1.29) 0.57 1.38 (1.01, 1.89) 0.04

Occupational group

Missing versus material/technical 1.17 (0.89, 1.55) 0.26 1.45 (0.99, 2.11) 0.06 0.85 (0.55, 1.31) 0.47

Humans versus material/technical 1.10 (0.85, 1.43) 0.46 1.38 (0.92, 2.08) 0.12 0.80 (0.55, 1.17) 0.25

Inform. versus material/technical 1.09 (0.80, 1.50) 0.57 1.72 (1.07, 2.77) 0.03 0.72 (0.46, 1.14) 0.16

123

Page 8: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

444 Int Arch Occup Environ Health (2012) 85:437–445

‘possibilities for development’ were signiWcant for men butnone were signiWcant for women. All psychosocial workenvironment factors showed stronger associations withretirement plans for men than for women. ‘Ageism’, in par-ticular, was a signiWcant predictor in the male subgroup, butnot at all in the female subgroup. This gender-speciWc resultwas not hypothesized in advance. The gender diVerencecannot be explained by a diVerent prevalence of poor psy-chosocial work environment at men’s and women’s work-places, since only small gender diVerences were seen forthe psychosocial work environment. Finally, we found thatpoor workability was a strong predictor of plans for earlyretirement for both men and women.

The results about workability and the gender diVerencein the psychosocial factors predictive value are in line withthe results from other studies. Workability has been foundto be a predictor of retirement thoughts and actual retire-ment (Sell et al. 2009; Tuomi et al. 2001). A Swedish studyfound that ageism was an issue for retirement plans for menbut not for women (Soidre 2005), and a Danish study foundsigniWcant gender diVerences in the psychosocial workenvironment’s eVect on planned retirement (Larsen 2008).

Since the gender diVerence is found in the eVect of ‘age-ism’ and not in the prevalence of ageism, we hypothesizethat a part of the explanation is that women on averageretire at a younger age than men—hence, they retire beforeageism becomes a notable issue. The reasons that womenretire early could be because in Denmark women generallyhave lower-paid jobs than men (Deding and Wong 2004).The gender diVerence in the result could also be due todiVerences related to the occupational group women andmen mostly are occupied in. For this reason, we controlledfor occupational group, vocational education and physicaljob strains in our analyses and the diVerence in the gender-stratiWed results remained. The gender diVerence could alsobe related to factors outside work, e.g. the presence ofgrand children or a retired (older) spouse (Jensen 2005).Another explanation could be that women have a less work-related self-concept (Alvesson and Due Billing 1997).These factors have not been included in our analyses.

The study uses cross-sectional data and plans of retire-ment instead of actual retirement. From a theoretical per-spective, retirement thoughts, retirement plans and actualretirement are not the same (Wang and Shultz 2010; Topaet al. 2009). However, in praxis, there is very high concor-dance between planned retirement and actual retirement(Larsen 2008). In a cross-sectional study, we have to con-sider the problem of causality. Poor psychosocial workenvironment may aVect retirement plans as we claim, but itis also possible that retirement plans aVect the perceivedpsychosocial work environment. A longitudinal study mayalso have disadvantages. Unknown changes occurring afterthe baseline measurement at the employee’s workplace or

in his or her health situation could aVect the actual retire-ment time years later.

Ageism is a form of discrimination, which can be deW-ned as unfavourable treatment of employees due to theirmembership of a particular groups, regardless of theiractual qualiWcation and performance (see Becker 1957;Phelps 1972). Six aspects of ageism at the workplace havebeen described in the literature: (1) reduced opportunitiesfor promotions, (2) less possibilities for training, (3) lessintroduction to new equipment, (4) less part in developmentappraisals, (5) less wage increase, (6) elderly workers notexpected to take part in change processes (Furunes andMykletun 2010). However, concerns for response burdendid not permit use of a multiitem scale, so we opted for aglobal question. We aimed to evaluate the attitude towardselderly employees in general rather than the perceived atti-tude towards the particular respondent. However, this glo-bal assessment may be inXuenced by the perceived attitudetowards the respondent him-/or herself. It is also possiblethat an elderly employee, if encountering ageism, will inter-nalize the attitude, lose conWdence and assess his or herworkability as low. In this case, our single item will under-estimate the true level of ageism.

Workability is closely associated with the health of theworker (Ilmarinen 2007; Martinez and Latorre 2006), andrecent studies have shown a beneWcial eVect on older work-ers’ health when they retire. After retiring, there was a sub-stantial improvement in SRH in the French GAZEL study(Westerlund et al. 2009) and a reduced mortality in a largeGerman study with data from a health insurance fond(Brockmann et al. 2009). Studies also exist that show thatsome retirees experience depressive symptoms and adecline of health after retirement (Wang and Shultz 2010).However, these results may partly be explained by selectiveprocesses that channels people with declining health intoretirement (Brockmann et al. 2009; Wang and Shultz2010). Furthermore, those who retire with a positive atti-tude in good health and good workability have a higherchance of a positive third age, than those there are pushedinto retirement (Tuomi et al. 2001; Wang and Shultz 2010).When the Danes are compared to other European National-ities, it is seen that they retire relatively late and that theyalso possess strong non-monetary reasons for wanting tohave a job (Jensen 2005).

The main result of this study is that poor psychosocialwork environment is associated with plans of retirementamong men. However, the main push factor from the labourmarket is for both genders reduced workability.

Acknowledgments We would like to thank Christian RoepstorV forthe data management of the study. The research was Wnanced bythe Danish Working Environment Research Fund, project number26-2009-03. The data collection for the DANES project was Wnancedby the National research centre of the working environment.

123

Page 9: The association between psychosocial work environment, attitudes towards older workers (ageism) and planned retirement

Int Arch Occup Environ Health (2012) 85:437–445 445

ConXict of interest The authors declare they have no conXict ofinterest.

References

Alvesson M, Due Billing Y (1997) Gender and work understandinggender and organization. Sage Publications, California, pp 54–81

Anonymous (2009) Forskerbeskyttelse (Researcher protection). http://www.dst.dk/TilSalg/Interview/Kunde/Forskerbeskyttelse.aspx.Statistic Denmark. Electronic Citation

Anonymous (2010) Pension og efterløn. https://www.borger.dk/Em-ner/pension-og-efterloen/Sider/default.aspx. Electronic Citation

Bech P, Rasmussen NA, Olsen LR, Noerholm V, Abildgaard W (2001)The sensitivity and speciWcity of the Major Depression Inventory,using the Present State Examination as the index of diagnosticvalidity. J AVect Disord 66:159–164

Becker GS (1957) The economics of discrimination. The University ofChicago Press, Chicago 60637

Bennington L (2001) Age discrimination: converging evidence fromfour australian studies. Empl Responsib Right J 13:125–134

Blekesaune M, Solem PE (2005) Working conditions and early retire-ment—a prospective study of retirement behavior. Research onAging 27:3–30

Brockmann H, Muller R, Helmert U (2009) Time to retire—time todie? A prospective cohort study of the eVects of early retirementon long-term survival. Soc Sci Med 69:160–164

Burr H, Bjorner JB, Kristensen TS, Tuchsen F, Bach E (2003) Trendsin the Danish work environment in 1990–2000 and their associa-tions with labor-force changes. Scand J Work Environ Health29:270–279

Deding M, Wong K (2004) Mænds og Kvinders løn: En analyse afløngabet 1997–2001. 04:10. Copenhagen, SFI Socialforskn-ingsinstituttet. Report

Elovainio M, Forma P, Kivimaki M, Sinervo T, Sutinen R, Laine M(2005) Job demands and job control as correlates of early retire-ment thoughts in Finnish social and health care employees. WorkStress 19:84–92

ESeC (2010) European socio-economic classiWcation (ESeC). http://www.iser.essex.ac.uk/research/esec. Electronic Citation

Furunes T, Mykletun RJ (2010) Age discrimination in the workplace:validation of the Nordic Age Discrimination Scale (NADS).Scand J Psychol 51:23–30

Gruber J, Wise D (1999) Social Security, retirement incentives, andretirement behavior: an international perspective. EBRI IssueBrief 209:1–22

Ilmarinen J (1997) Aging and work—coping with strengths and weak-nesses. Scand J Work Environ Health 23:3–5

Ilmarinen J (2006) The ageing workforce—challenges for occupa-tional health. Occup Med 56:362–364

Ilmarinen J (2007) The work ability index (WAI). Occup Med 57:160Jensen PH (2005) Reversing the trend from “early” to “late” exit: push,

pull and jump revisited in a Danish context. Geneva Pap RiskInsur Issues Pract 30:656–673

Kessler RC et al (2003) The world health organization health andwork performance questionnaire (HPQ). J Occup Environ Med45:156–174

Landis JR, Koch GG (1977) Measurement of observer agreement forcategorical data. Biometrics 33:159–174

Larsen M (2008) Does quality of work life aVect men and women’sretirement planning diVerently? Appl Res Qual Life 3:23–42

Loretto W, White P (2006) Population ageing and older workers:employers’ perceptions, attitudes and policies. Popul Space Place12:341–352

Lund T, Borg V (1999) Work environment and self-rated health as pre-dictors of remaining in work 5 years later among Danish employ-ees 35–59 years of age. Exp Aging Res 25:429–434

Lund T, Villadsen E (2005) Who retires early and why? Determinantsof early retirement pension among Danish employees 57–62years. Eur J Ageing 2:275–280

Martinez MC, Latorre MDDD (2006) Health and work ability amongoYce workers. Rev Saude Publica 40:851–858

Nielsen J (1999) Employability and workability among Danishemployees. Exp Aging Res 25:393–397

Nunally JC, Bernstein IH (1994) Psychometric Theory. McGraw-Hill,New York

Pejtersen JH, Kristensen TS, Borg V, Bjorner JB (2010) The secondversion of the copenhagen psychosocial questionnaire. Scand JPublic Health 38:8–24

Phelps ES (1972) Statistical theory of racism and sexism. Am EconRev 62:659–661

Riach PA, Rich J (2006) An experimental investigation of age discrim-ination in the French labour market. IZA Discussion Papers

Riach PA, Rich J (2007a) An experimental investigation of age dis-crimination in the English labor market. IZA Discussion Papers

Riach PA, Rich J (2007b) An experimental investigation of age dis-crimination in the Spanish labour market. IZA Discussion Papers

Schalk R (2004) Changes in the employment relation across time. 304.United States, Oxford University Press. The employment rela-tionship: examining psychological and contextual perspectives.Serial (Book, Monograph)

Seitsamo J (2005) Qualities of work, functioning and early retirement.A longitudinal study among Finnish ageing workers in 1981–1997. Int Congr Ser 1280:136–141

Sell L, Bultmann U, Rugulies R, Villadsen E, Faber A, Søgaard K(2009) Predicting long-term sickness absence and early retire-ment pension from self-reported work ability. Int Arch OccupEnviron Health 82:1133–1138

Sigg R, De-Luigi V (2007) The success of policies aimed at extendingworking life. In: Development and trends: supporting dynamic socialsecurity. World social security forum 29th ISSA general assembly,International Social Security Association, Moscow, pp 51–60. Report

Soidre T (2005) Retirement-age preferences of women and men aged55–64 years in Sweden. Ageing Soc 25:943–963

Sutinen R, Kivimaki M, Elovainio M, Forma P (2005) Associationsbetween stress at work and attitudes towards retirement in hospitalphysicians. Work Stress 19:177–185

Topa G, Moriano JA, Depolo M, Alcover CM, Morales JF (2009) Ante-cedents and consequences of retirement planning and decision-making: a meta-analysis and model. J Vocat Behav 75:38–55

Townsend P (2006) Policies for the aged in the 21st century: more‘structured dependency’ or the realisation of human rights? AgeingSoc 26:161–179

Tuomi K, Ilmarinen J, Seitsamo J, Huuhtanen P, Martikainen R,Nygard CH, Klockars M (1997) Summary of the Finnish researchproject (1981–1992) to promote the health and work ability ofaging workers. Scand J Work Environ Health 23:66–71

Tuomi K, Huuhtanen P, Nykyri E, Ilmarinen J (2001) Promotion ofwork ability, the quality of work and retirement. Occup Med51:318–324

van Oorschot W, Jensen PH (2009) Early retirement diVerences be-tween Denmark and The Netherlands. A cross-national compari-son of push and pull factors in two small European welfare states.J Aging Stud 23:267–278

Wang M, Shultz KS (2010) Employee retirement: a review and recom-mendations for future investigation. J Manag 36:172–206

Westerlund H et al (2009) Self-rated health before and after retirementin France (GAZEL): a cohort study. Lancet 374:1889–1896

123