Satisfaction with Work in a European Perspective: Center and Periphery, "Old" and "New" Market Economies Compared

SATISFACTION WITH WORK IN A EUROPEAN

PERSPECTIVE: CENTER AND PERIPHERY, “OLD”

AND “NEW” MARKET ECONOMIES COMPARED*

Márton MEDGYESI–Péter RÓBERT

TARKI Social Research Centre

Budapest, Budaörsi út 45. H-1112; e-mail: [email protected]

Institute of Sociology, ELTE University

Budapest, Pázmány Péter sétány 1. H-1117; e-mail: [email protected]

Abstract: The analysis approaches satisfaction with work as affected by demographic and

social characteristics as well as by labor market and labor organization position.

Furthermore the impact of expectations to and evaluation of the job is also considered. The

international variation in the explanatory mechanisms of satisfaction with work is analyzed

by comparing five groups of European societies. First, distinction is made between ‘old’

European market economies and ‘new’ post-socialist countries. Second, based on economic

indicators, the first group of countries is divided into three subgroups: West (Center 1),

Scandinavia (Center 2), and Periphery, while the second group of countries is divided into

more developed and less developed transition societies. The ISSP 1997 Work Orientation

Module data are used for the analysis. Only respondents in labor force are analyzed

(N=11739). The paper presents descriptive statistics for the clusters of the countries as well

as for the dependent and independent variables in the analysis. Then, ordered logit models

are used to predicting satisfaction with work. The explanatory variables contain objective

status indicators, subjective evaluation of the job, the country groups and interaction terms.

Results reveal that both status indicators and attitudes toward the job are significant

predictors of the general satisfaction in agreement with gender paradox, life cycle,

reference-group and status discrepancy hypotheses. However, these explanatory

mechanisms vary a lot by the groups of countries. If controlling for composition effects

within these groups of countries, Scandinavia turns out to be a place with highest satisfaction

and developed transition societies are characterized by the lowest satisfaction with work.

Keywords: labor market, satisfaction, post-socialist countries, periphery

1417-8648/$ 20.00 © 2003 Akadémiai Kiadó, Budapest

Review of Sociology Vol. 9 (2003) 1, 43–68

* This is a revised version of the paper presented at the Session 14, ISA RC20 on Comparative Sociology,

at the XV ISA World Congress of Sociology, Brisbane, Australia, July 7-13, 2002.

INTRODUCTION: WHY TO STUDY SATISFACTION WITH WORK?

The importance of studying work attitudes is based on their close connection tovarious fields of observed behavior (Hakim 1991, Clark 1997). The main argumentbehind the research on this subject is that satisfied workers do more and better job thanless satisfied ones. Several further assumptions stress the relation between jobsatisfaction and behavior. They assert that unsatisfied workers are more likely to changejob, to be absent from work, to spend more time with extra-work activities and earn morefrom these activities instead from their main job. Literature also draws attention to thefact that satisfaction with work is a central component of the quality of life as well andhas an impact on mental health (Curtice 1993). Investigations related to social structuretreat work conditions as an important factor. Hradil (1994) underlines the importance offour work-related fields: job security, flexible work hours, work conditions, andenvironment of work (= satisfaction with work).Satisfaction with work is a centralcomponent of work attitudes. In general, work attitudes depend on work values and workattributes. Satisfaction with work is an outcome of workers’ perception about whathis/her job is and of workers’ opinion about what his/her job ought to be. However,satisfaction with work is influenced by a great number of other factors in addition tothese two components. It is affected by the organizational and institutional context ofwork relations as well as by the individuals’ socio-economic background, which alsoplay an obvious role in influencing what kind of job one intend to achieve.

In this paper, various determinants of satisfaction with work are investigated. Part

of them relates to the individuals’ socio-demographic characteristics as well as to their

position in the division of labor and in the work organization. Another set of

explanatory variables contains measures about attitudes towards the important

characteristics of a job as well as opinions about individuals’ present job. The main

focus of the analysis, however, is the international variation in the explanatory

mechanisms of satisfaction with work.

In the next sections, we go through the explanatory mechanisms first, referring tothe existing theory and hypotheses. Second, the perspective of internationalcomparison is set. Then, data and methods are shown. Afterwards, we present thefindings of the empirical analysis. The paper ends with a concluding section.

EXPLANATORY MECHANISMS OF SATISFACTION WITH WORK

This section is devoted to the mechanisms that represent major correlates of work

values and, consequently, may influence satisfaction with work (Kalleberg 1977). Based

on the literature and on our previous work with this data (Medgyesi–Róbert 1998;

Róbert–Medgyesi 1999), we intend to consider 1. demographic characteristics like

gender or age; 2. status attributes and features of labor market capability; 3. position in

the work organization; and 4. expectations to and evaluation of the job with special focus

on discrepancy between these two attitudes. Discussion of the explanatory mechanisms

leads to the hypotheses on the international variation in these characteristics.

Review of Sociology 9 (2003)

44 MÁRTON MEDGYESI–PÉTER RÓBERT

Demographic characteristics

Previous research on the topic found a seemingly paradoxical relation betweengender and satisfaction with work. Although women generally have not as good workconditions as men have (lower salaries, less promotion opportunities, lower jobsecurity), on the average they tend to be more satisfied with work (Hakim 1991;Curtice 1993; Clark 1997). The cause of this ‘gender paradox’ can be that men andwomen have different priorities when evaluating their job. For women, work is only asecondary source of self-esteem while family and housework play a more importantrole in their life (Polachek 1981). Different work values as well as the mechanism ofself-selection based on these orientations have an impact on satisfaction. Thisassumption is supported by some of the interesting findings of the previous research,namely that women who work in part-time jobs are especially satisfied with work(Curtice 1993). At the same time, previous research did not find substantial differencein work satisfaction for men and women if both work in a full-time job.Kalleberg andLoscocco (1983) investigated the impact of age on work satisfaction. They found thatage affects work conditions and work values, and both cohort effects and life cycleeffects are present. The explanation of life-cycle effect is that older people are at amore advanced stage of their career and consequently can achieve better jobs thanyounger individuals. The cohort effect can come from the fact that members of largercohorts have more difficulties to find a good job as compared to members of smallercohorts. The life cycle effect has another feature, namely that the importance ofspecific job characteristics can also change as individuals get older. The samephenomenon from a cohort perspective means that socialization to work of differentcohorts can differ and it is possible that members of subsequent cohorts evaluatedifferently the same or similar jobs. Since studies based on cross-sectional data haveno proper opportunities to separate cohort and life-cycle effects, they tend to find theimpact of the age as being curvilinear (e.g. Clark et al. 1995; Blanchflower–Oswald1999).

H1. We expect that gender will have a main effect on satisfaction with work and

men will be less satisfied than women.

H2. We basically expect that older respondents will be more satisfied with work.

However, we do not expect this relationship to be necessarily linear. Thus, we intend to

test the curvilinear effect of age, too.

Status attributes, labor market capability

Human capital theories assert that the level of education influences the returns with

respect to job and income, the individuals can realize. According to Becker (1975)

people invest in their human capital in order to achieve better jobs with higher

occupational prestige and higher income and, in this way, higher general social status.

Consequently higher educated people are expected to have better jobs. This does not

mean, however, that better educated people will be more satisfied with their work

because they also tend to have higher level of aspiration. Moreover, people with higher

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 45

education and in higher social status are frequently less satisfied with their work

conditions because they tend to compare them to the conditions of those who have

even better jobs. Reference group theory (Merton 1968) provides a framework for

interpretation here and this concept argues that people in higher status compare

themselves frequently to such ‘circles’ occupying positions above them.

In addition to the impact of education, we intend to investigate the influence of

another relevant attribute of social status, namely income. Although when returns to

human capital investments are investigated in the economic literature, achieved

income is regarded as the main measure, this relationship between education and

income is influenced by several other structural elements of the labor market

(Farkas–England–Barton 1988; Beck–Colclough 1988). The effect of income cannot

be, consequently, derived simply from human capital investments but it has a more or

less independent influence on satisfaction with job.

More regular or more causal participation in work are good indicators of labor

market embeddedness. Full-time workers or part-time (temporary) workers have

different relationship to job, work plays a different role in their life, as mentioned

above when gender differences in satisfaction with work was discussed. However, as

employment relations are getting more flexible in modern societies including temporal

variances of working activities as well as various types of contracts

(Schömann–Rogowski–Kruppe 1998; Hakim 1998), labor market capabilities tend to

contribute stronger to satisfaction with work. (This is discussed in more details in the

next section, too.)

H3. We assume that respondents with higher status attributes and stronger

embeddedness in the labor market are more satisfied with work. However, we do not

expect this relationship to be necessarily linear, especially in the case of the level of

education.

Position in work organization and employment status

Organizational and institutional contexts also play an important role in influencingwork attitudes and satisfaction with work. The mixed model by Kalleberg and Reve(1992), building on sociological and economical approaches, focuses on employmentcontracts. Employment relations have an increasing importance in the sociologicalliterature arguing that modern societies are post-industrial service economies where aservice class is emerging and employment relations provide the basic characteristics ofthis class (Goldthorpe 1982). In fact, employment contracts with different duration,different guarantees for job security or different features of organizationaladvancement tend to form the actual basis for the new type of employment relations.The employment contracts mirror authority and dependence relations betweenemployers and employees. The managers and supervisors have probably the bestpossibilities for more advantageous employment contracts. Satisfaction with work wasfound higher for those in supervisor position in previous analyses, too (e.g.Blanchflower–Oswald 1999).

Review of Sociology 9 (2003)

46 MÁRTON MEDGYESI–PÉTER RÓBERT

Sociological theories of contracts assert that exchange relations on the labor market

are usually unequal and the results of exchanges depend on the initial distribution of

resources. Those employees who have less personal assets and labor market

capabilities can decide to form coalitions and co-operate in defending their interests.

Workers typically join to or form unions where they can negotiate collectively on the

terms of their employment contracts. Thus, unionism – if it is strong and provides

support for its members – can also contribute to better work conditions and to higher

satisfaction with work. On the other hand, union members as individuals are laborers

with less assets and lower social status. If so, in line with our hypothesis on status

attributes, union membership can affect negatively satisfaction with work. This was

the finding of previous studies (Freeman 1978; Borjas 1979).

As mentioned above, employment contract is an important factor for the emerging

service class. This service class is a fraction of the middle class in the society, the

so-called ‘new’ middle class. In addition, the ‘old’ middle class, the petty bourgeoisie

is another essential fraction of the middle class. Self-employment constitutes a special

position in the labor market, a position without employment contract. Self-employment is a decision aiming at larger independence, personal and professionalfreedom, larger prospects for occupational and financial career, etc. Self-employmentalso provides more freedom of deciding about work conditions, larger degree of workautonomy, which can have a positive effect on satisfaction with work. At the sametime, self-employment can be a consequence of a structural constraint in the labormarket or in the individual assets of the person when somebody is not able to find a joband to be part of the labor force as an employee. Previous evidence onself-employment suggests that this position increases satisfaction with work(Blanchflower–Oswald 1999).

H4. We expect that supervisor position and self-employed position will have a

positive impact on satisfaction with work. The effect of union membership can be

either positive or negative.

Expectations to and evaluation of the job: discrepancy of job attributes

People usually have certain expectations to the job. E.g. the job should be secure,

interesting and independent, or the job should provide good salary and possibilities for

advancement. These features can be more important or less important for different

workers. Expectations to the job have an effect on satisfaction with work. At the same

time, people evaluate their job in accordance with their expectations. The evaluation of

the job is usually based on the same characteristics: security, excitement, independence,

high salary, and good possibilities for advancement. People are more satisfied with their

job if it meets their wishes. In sum, the larger the expectation to work attributes, the

lower the satisfaction with work. (It is always more difficult to meet higher demands.)

And the better the evaluation of work attributes, the higher the satisfaction with work.

However, even if a job does not meet certain requirements, e.g. it is not interesting

or independent or does not provide good chances for advancement, this does not

necessary mean that an individual will not be satisfied with work. Satisfaction with

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 47

work is connected to the existence or non-existence of job attributes that are important

for someone. If the job is not interesting but the person does not have such

expectations, he/she may be satisfied with his/ her work. However, if the job is not

interesting and the person considers this attribute as an important feature of the job,

he/she will not be satisfied with his/her work. This is what we call discrepancy in the

job attributes and this is what matters for satisfaction with work.

Furthermore, there are other subjective aspects of work conditions that are worth to

take into account as correlates of job satisfaction. These are how demanding the job is

physically or how stressful the job is mentally. These conditions can influence

satisfaction with work negatively. The good work related climate and the good

personal relations to colleagues and to managers and supervisors at the workplace,

however, have positive impact on satisfaction with work.

H5. The higher the discrepancy one experiences with job attributes, the lower the

satisfaction with work. The better the evaluation of the work conditions, the higher the

satisfaction with work.

INTERNATIONAL VARIATION IN THE LEVEL AND IN THE

EXPLANATORY MECHANISMS OF SATISFACTION WITH WORK

This research focuses on international variation in satisfaction with work. We intendto compare the level as well as the determination of satisfaction with work in variousEuropean countries. The data set we employ for this paper contains non-Europeancountries as well but we decided to exclude them from the present analysis because ourresearch interest focuses on the differences between traditional market economies and‘new democracies’ (former socialist countries) as well as between countries in centerand peripheral position within Europe. (An obviously relevant next step can be includingfurther countries where data are available and divide them according to their center orperipheral position. But this step has been postponed.)

The strategy, however, is not to investigate and compare different separatecountries one by one. Nations, for which data are available, were put together intoclusters based on macro level information we describe later in the measurementsection. The reason that we carry out our comparison on this more aggregate level isthat these groups of countries represent characteristic types in Europe while thisstatement holds true less for the individual nations within the groups. The countries wehave in our data file cannot be considered as a result of any sampling strategy from theEuropean societies, it is quite accidental that data are available for them and notavailable for other countries. Nevertheless, we cannot defend a claim that our groupsare not homogenous with respect to our research question. Certainly there aredifferences among the nations within each group.1 However, these differences amongthe individual countries, if we have a relatively large number of them, are frequentlyvery difficult if not impossible to explain. Using measurement terms, the data from

Review of Sociology 9 (2003)

48 MÁRTON MEDGYESI–PÉTER RÓBERT

1 E.g. Vecernik (2000) points out that satisfaction with work is in high contrast in the Czech Republic andHungary; respondents are more satisfied in the former and less satisfied in the latter country. (See alsoour data in Figure A1 in the Appendix.)

each country has larger measurement errors as compared to data from aggregategroups of the same countries. Thus, we expect more robust results from ourcomparison by using this research design. As we will show the combination of thecountries into five groups is not ad hoc, it is based on macro economic data as well inaddition to political and economic geography.

The analysis will basically refer to four types of European societies, traditional

market economies and transitional (former socialist) countries, which are either in

central or in peripheral position. Satisfaction with work is regarded an attitude item.

Previous international comparative research (e.g. Eurobarometer) provided evidence

that different attitude questions usually reveal lower level of satisfaction in the former

socialist countries than in the market economies. Moreover, work attributes

influencing satisfaction with work are probably less favorable in the peripheral

countries than in the countries in central position. Accordingly, we assume a kind of

hierarchical ‘order’ of these country groups where traditional market economies in

central position are on the top and peripheral transition societies are on the bottom.

H6A. We expect that satisfaction with work will be lower in the former socialist

countries and it will be the lowest in those being less developed among them.

Respondents in market economies will be more satisfied with their work, especially

those who live in countries that occupy a central position.

An alternate hypothesis can be formulated on the ground of the reference grouptheory. In original form, the theory implies that people’s judgments are based onexperiences they have with friends, co-workers, neighbors in addition to their ownexperiences (Merton 1968) and these judgments consider less the whole society orsuch segments of the society people have no personal experiences with. However, asmedia and modern techniques bring the whole world closer to the people, they will geta view about other parts of the society, about other countries and it increases thepossibilities for comparison and expands the possible groups of reference. This is morethe case for those people who are in higher social status and have better access to thesemodern techniques. The original form of the concept, namely, that individuals seemostly others being similar to them and, consequently, will think that the whole worldis similar to their personal world, holds less – especially to those with higher socialstatus. A next, psychological assumption is that people, if making their judgments onthe ground of comparisons, will compare themselves upward, to those who have betterconditions rather than downward. In consequence, individuals tend to see themselvesas occupying middle or lower positions in the society (or in any social hierarchy), evenif being in relatively good status in reality (Lopreato–Hazelrigg 1972). On a moreaggregate level, we also anticipate that this difference in having narrower or widerhorizons for reference groups can exist on the ‘average’ level of societies in addition tosocial groups.2 This leads us to propose the alternate hypothesis that people in thosesocieties being not in the worse but also not in the best position with respect to the

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 49

2 On this ground the explanatory mechanism has been developed why Hungarians were more ‘unhappy’ in

the first part of 1990s in comparison to people in other countries. On the one hand, they had a better

reference from the socialist past, when Hungary provided better living conditions than any other former

socialist country. On the other hand, they compared themselves to the Western market economies with

higher living standards (Róbert 1999).

determinants of satisfaction with work can be the least satisfied with their situation.We argue that this happens because they compare their circumstances to those who areabove them, who have better work attributes and this creates higher tension.

H6B. We assume that satisfaction with work will be lower in the developed

transition societies as compared to the other groups of nations.

The explanatory mechanisms behind satisfaction with work, the determinants of this

attitude may also differ for the different group of countries. The ‘gender effect’ can

depend on the level of female participation in work or on women’s part-time

participation in the labor force. In the former socialist countries the level of full-time

employment of women was high, though it dropped substantially in the 1990s (Róbert

1998), but part-time employment did not increase (Drobnic 1997). Career advancement

under socialism was highly age dependent; better work conditions, higher salaries were

stronger connected to older age in the former socialist ‘gerontocracies’. However, this

influence was definitely crosscut by transformation of the system when a rapid change of

generations took place in these countries (for Hungary see Kolosi–Sági 1998).

As far as education is concerned, its contribution to status allocation was markedlydifferent in the former socialist countries and in the market economies. Two kinds ofdifferences can be underlined and they go into different directions. On the one hand,scholars of status attainment analyses found that the influence of education onoccupation is stronger under socialism (e.g. Simkus 1981). On the other hand, incomeinequalities were traditionally lower under socialism as compared to the marketeconomies. Following an equality principle, the returns to human capital investmentswere lower in the socialist countries. This latter relationship, however, has become asubject of change in the transition economies of the 1990s.3 These contradictingprocesses make it difficult to predict how education will influence satisfaction withwork in the different groups of countries. For income, we prefer to assume thatfinancial rewards will matter stronger in the transition economies. Though Inglehart(1990) claimed that post-materialism was on the increase even under socialism, webelieve that the materialist orientation is still stronger in the transition countries than inthe traditional market economies. Consequently, level of income is expected to affectsatisfaction with work stronger in the former socialist countries.

Supervisor position and self-employment may influence satisfaction with work

differently in the various groups of nations. Considering the traditionally higher

significance of power in attributing social status under socialism, supervisor position

can be a stronger predictor of satisfaction with work in the transition economies.

However, if the reference group theory holds, supervisors – even if they have stronger

advantages in salaries, work conditions, etc. – may be less satisfied with their work in

the former socialist countries. The group of self-employed is rather heterogeneous and

this makes difficult to foresee how its effect will vary.

Trade union membership has also quite different social meaning in the various

groups of nations. This has more relevance in the traditional market economies where

entry into the trade union is based on the consideration and expectation that membership

will increase job safety, improve work conditions, etc. In this respect, there are marked

Review of Sociology 9 (2003)

50 MÁRTON MEDGYESI–PÉTER RÓBERT

3 See e.g. Kertesi–Köllõ (2002) for Hungary or Vecernik (1999) for Czech Republic.

traditional differences between the Nordic countries or Germany with stronger

unionism, on the one hand and the Southern European countries with less influence of

the trade unions, on the other. Trade union membership was quasi automatic under

socialism and it had no real advantage for the members. Poland can, however, be

mentioned as an obviously different case. But trade union membership based on a free

choice has strongly dropped in the 1990s in most of the transition economies.

Finally, we assume that discrepancy between expectations toward the job and the

exact job characteristics will have varying effects on the satisfaction with work in the

different groups of countries. In line with the assumption of stronger materialist

attitudes, discrepancy in high pay may be more important for people in the transition

economies. The same holds for the discrepancy in job security because the decrease of

this type of safety was shocking experience for workers in the former socialist

countries when unemployment increased in the 1990s. Previous research evidence

seems to support this postulation, namely that work values referring to high pay or

secure job are more important in the post-socialist countries (Ester et al. 1994). On the

other hand, post-materialist aspects of the job like discrepancy in interesting or

independent work may have weaker effect on satisfaction with work in the transition

countries than in the traditional market economies.

DATA, VARIABLES AND METHODS

Data are taken from the ISSP 1997 module on Work Orientation. We have selected

16 countries where all independent variables we intended to use in the analysis were

available (Bulgaria, Czech Republic, Denmark, France, Germany, Great Britain,

Hungary, Italy, Norway, Poland, Portugal, Russia, Slovenia, Spain, Sweden,

Switzerland) with a total N of 11739, only respondents in labor force.

The dependent variable of the paper (satisfaction with work) was measured on a7-point scale with the following categories: 1. completely satisfied; 2. very satisfied; 3.fairly satisfied; 4. neither satisfied nor dissatisfied; 5. fairly dissatisfied; 6. verydissatisfied; and 7. completely dissatisfied. This scale has been later reversed in orderto get positive estimates in the multivariate analysis. Figure A1 in the Appendixdisplays the distribution of the dependent variable for the 16 countries. Since thedistribution is not equal among the categories, the lower three categories(dissatisfaction) and the upper two categories (satisfaction) are combined in the figure.

The crucial point of the analysis is the grouping of the selected countries into clusters.

This is based on three pieces of information, partly on macro level: GDP per capita; Gini

coefficient of households’ net income; and proportion of trade union members. Table A1 inthe Appendix provides the appropriate figures in this respect. GDP per capita is measuredin USD, at comparative purchasing power parity. This measure ranges from Norway(highest) to Bulgaria (lowest). We employ this measure in line with the assumption that ahigher level of economic development results in higher level of wages and of consumptionand this leads to higher level of satisfaction with work in the end. The Gini coefficientinforms about income inequalities in the societies and displays the social distance amongthe groups in the labor force. This is also meaningful from the viewpoint of the reference

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 51

group theory mentioned earlier in the theoretical section of the paper. In this respect,Denmark is on the top and Russia is at the bottom of the hierarchy. Satisfaction with workis expected to be lower in countries with higher income inequalities. Finally,(de)regulation of the labor market is important for characterizing the institutional structureof the labor market. Elements like employment protection or bargaining on wage levelsand wage structures can be considered here. For the latter one, three features of the labormarket are vital: how large fraction of the labor force is affected by wage bargaining(expansion); how centralized or decentralized is the level of wage bargaining(centralization: country level, industry level, company level); how large is the consensusamong the parties in the wage bargaining (degree of coordination). Only one piece ofinformation on the first feature was available for all 16 countries and this is the percentageof trade union members. (In fact, even this is taken from the data we use in the paper andnot from published macro sources.) This information is regarded as a measure beingrelevant both for wage bargaining and for employment protection. Denmark is again onthe top of this hierarchy and Spain is at the bottom. We employ this measure in line withthe assumption that satisfaction with work varies with the strength of unionism or with thestatus attributes with the union members.4

We intended to produce groups of the countries by combining the traditionalmarket character, the post-socialist feature, and the central or peripheral position ofthem. The macro level measures provide a solid support for these clusters. E.g.Bulgaria or Russia deviates quite clearly from the other transition societies. Withrespect to the market economies, Portugal and Spain seem to differ from the rest.Furthermore, the three Scandinavian countries tend to form a separate unit as well.Consequently, five groups of the 16 nations were separated: developed West Europeancountries; developed North European countries; less developed market economies inthe European periphery; developed (central) transition societies; and less developed(peripheral) transition societies. (See Table 1)

Table 1. Definition of country-groups

Country- group Countries

Western Europe France, Germany, Great Britain, Italy,Switzerland

Nordic countries Denmark, Norway, Sweden

EU periphery Portugal, Spain

Developed transition countries Czech Republic, Hungary, Poland, Slovenia

Less developed transition countries Bulgaria, Russia

Review of Sociology 9 (2003)

52 MÁRTON MEDGYESI–PÉTER RÓBERT

4 Perhaps it is worth underlining once more that we did not follow the usual practice of the similar analysesof ISSP data. This would have meant clustering the countries on the basis of the dependent variable, i.e.satisfaction with work and then trying to explain why different countries ended up in the same group.This is frequently an impossible task, though this routine produces homogeneous groups of countries in apractical but not really conceptual sense.

Turning to the measurement of the independent variables, we have social-demographic

characteristics on the one hand and attitude variables on the other, which are expected to

affect satisfaction with work. The first bunch of the independent variables contains gender

(women = 1); age measured in years + its quadratic term; level of education measured in

years + its quadratic term; income (three terciles were computed; the lowest tercile is the

reference category while the middle and the highest terciles = 1); position in work

organization (supervisor = 1); self-employment (self-employed = 1); temporary work

(part-timer = 1); and union membership (member = 1).

For the independent attitude variables, five discrepancy measures were computed

with respect to job security, high pay, good advancement in the hierarchy at the

workplace, level of interestingness of the job and independence in the job. In the

survey, respondents had to express first on a 5-point scale how important these five job

characteristics are for them when they consider a job. Second, they had to evaluate on a

similar 5-point scale how these five job characteristics are attributable to their present

job. It was regarded a discrepancy between expectations and reality, if the respondent

evaluated a given feature of his/her job by 2 points lower as compared to the

importance of the same job attribute. These cases were coded as 1 for the five job

characteristics separately.

Furthermore, respondents had to evaluate their work on a 5-point scale whether it ishard physically or stressful mentally. Similarly, they had to evaluate whether havinggood relations to their colleagues and supervisors at the workplace. These answersserved as a basis for further three dummy measures where respondents were coded as 1if these features related to their work. Descriptive statistics for all independentvariables described here are presented in Table A2 in the Appendix.

In addition to the descriptive information on the bivariate relationship between thedependent and the independent variables we intend to apply multivariate technique aswell in order to see how satisfaction with work can be predicted by the independentvariables. Satisfaction with work is measured on a scale, which is definitely ordinal butthis level of measurement is not perfect for using OLS regression for the purpose ofprediction. Instead, we will apply the ordered logit model. This method assumes thatthe dependent variable is a latent one and is measured on an interval scale but onlycertain values of the scale (from 1 to 7) are empirically observed. Then, the statisticalanalysis provides estimates for the predictor variables using maximum likelihood(ML) method and, in addition, provides estimates for the (six) cut-points between the(seven) categories of the dependent variable. The model assumes that the increase inthe satisfaction with work cannot be observed (it is a latent variable) to a certain point.But when the increase reaches a cut-point it becomes observable (Greene 1993; Long1997).5

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 53

5 We have a structural and a measurement model. The structural equation is yi*=xi �+ �i, where yi* is the

latent dependent variable with values between ±�, and xi is the matrix for the independent variables,

symbol � is the vector of the parameters, �i is the residual assumed to have a logistic distribution. Themeasurement equation is yi=m and �m-1 � yi* <�m where yi is the observed dependent variable with valuesm=1…j (in this case from 1 to 7) and the �m are cutpoints (estimated by the statistical procedure) dividingthe continuous yi* variable into m ordered categories. In a previous paper, Blachflower–Oswald (1999)applied the same statistical method for predicting job satisfaction.

BIVARIATE RELATIONSHIPS BETWEEN SATISFACTION

WITH WORK AND ITS CORRELATES

We start our analysis with presenting descriptive findings on satisfaction with workfor the five groups of countries. First, we display the differences in the mean level ofsatisfaction with work by level of education (Figure 1). In most of the country clustersthe mean value of years of schooling was around or over 11 years (with the exceptionof Portugal and Spain, the peripheral market economies). This is why we divided therespondents into three groups; those with maximum 11 years of education (a lowerlevel of schooling than the average and – in most cases – an education withoutcompleting secondary level which usually requires 12 years); those with 12-15 yearsof schooling (secondary or at least lower tertiary education); and those who spent 16years or more in school (high tertiary education and post-graduate training). Iffocusing on the first and third categories, we found that education did not differentiatefor the two groups of the developed market economies; this difference is moderate forthe peripheral market economies; but it is more marked for the transition societieswhere respondents with higher level of education are more satisfied with work.

Figure 1. Satisfaction with work by years of schooling in country-groups (scale averages)

Income seems to discriminate stronger than education (Figure 2). Here wedisplayed the mean level of satisfaction with work by income focusing on the lowestand the highest terciles. Differences are smaller for those who live in the developedmarket economies, though higher earnings go with higher satisfaction with work inthese countries as well. In fact, money seems to buy more satisfaction in theScandinavian countries. Data indicate larger variation for the peripheral than thecentral market economies. But income discriminates much stronger for the transitionsocieties and especially for the peripheral ones. In Bulgaria and Russia we can observeespecially large difference in the mean level of satisfaction with work connected to therespondents’ better income position.

Review of Sociology 9 (2003)

54 MÁRTON MEDGYESI–PÉTER RÓBERT

Figure 2. Satisfaction with work by income terciles in country groups (scale averages)

With respect to position in the work organization, our results indicate that executives

and supervisors are usually more satisfied with their work, as we expected, but there is an

obvious variation for the groups of countries, similar to those we have seen before. The

supervisor – subordinate cleavage discriminates in satisfaction with work less for the

developed market economies; a stronger difference can be observed for the peripheral

market economies and the developed transition societies; while being in supervisory

position matters most for those living in the peripheral transition countries.

Employment status on the labor market, being an employee or self-employed, also

discriminates in satisfaction with work. In line with the assumptions, self-employed

turned out to be more satisfied than employees were. In this case, the employee –

self-employed position seems to make the smallest difference for the peripheral

market economies. A relatively higher level of satisfaction with work is connected to

self-employment in the developed market economies; this is more the case in Western

Europe and less the case in the Scandinavian countries. Self-employed are more

satisfied with work in the post-socialist countries as well but the difference is not much

for the more developed transition economies, it is apparently smaller than the

dissimilarity in many developed market economies with the exception of the

Scandinavian ones. The employee – self-employed position seems to have the largest

discriminative power in the peripheral transition societies where self-employed are

much more satisfied with work as compared to employees.

In addition to the bivariate relationship between the status variables and

satisfaction with work, we looked at the connection to the subjective measures as well.

We would like to stress two kinds of discrepancies: the one that relates to interesting

work and the one that relates to high pay. The first one represents a discrepancy of

post-material kind, the case when the work should be interesting but the present job is

not interesting; the other corresponds to the material kind of discrepancy, when the

work should be a well paid one but the present job does not pay well. In the light of the

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 55

results, it is obvious that discrepancy decreases satisfaction with work and it has

country-group variation. Discrepancy with regard to interesting work discriminates

stronger for the Nordic country group than for the Western market economies. There

seem to be a rather small difference between the cluster of the peripheral market

economies and the cluster of the developed transition societies. Discrepancy for

interesting work discriminates less for the peripheral post-socialist countries. On the

contrary, discrepancy with regard to high pay makes the largest distinction for this

group of countries. For the other four clusters, the scale averages indicate a rather

similar picture with respect to the differences between the means of work satisfaction.

PREDICTING SATISFACTION WITH WORK

IN A MULTIVARIATE PERSPECTIVE

The ordered logit analysis is performed in four steps. Model 1 contains the objectiveand subjective variables that relate to the respondents’ social status and demographiccharacteristics as well as to his/her opinion about his/her job. This model also contains adummy variable separating only the market economies (3 clusters) and the transitionsocieties (2 clusters). We present the estimates for the transition societies in comparisonto the market economies as a reference category to get a general view about the majorcleavage. Model 2 adds interaction terms between the market economy – transitionsociety dummy and other independent variables in order to see the variation in the effectsof the predictor variables. Only the interactions with significant estimates are shown.These models are displayed in Table 2. In Model 3 the market economy – transitionsociety dummy is replaced by the dummies for the separate country clusters in order toperform the comparison in a more elaborated manner in line with the goals of the paper.Here we have four dummies and the developed (central) West European country group isused as a reference category. Finally, in Model 4 new interaction terms are added to theprevious equation in order to get a better insight how the influence on satisfaction withwork varies for the different clusters of countries. We present only those interactionswith significant estimates. These models are displayed in Table 3.

Estimates of Model 1 confirm a large part of our assumptions. As expected, womenare more satisfied with work; higher earnings increase satisfaction with work; peoplein supervisory position and self-employed are also more satisfied with work. Theeffect of education is, however, not linear. Higher level of schooling decreases thesatisfaction with work (negative coefficient for the main effect) but this influencedecreases and turns back as the level of education increases further (positive effect forthe quadratic term of years of schooling). This means that those with a middle level ofschooling are the less satisfied with work – if the effect of education is controlled forother independent variables in the model. (This pattern deviates from the one we havepresented in the previous section when the descriptive bivariate relationships wereshown.) This result seems to be closer to the assumption derived from the referencecategory theory. Other status characteristics like age, part-time work or trade unionmembership do not seem to be significant predictors of satisfaction with work in ourmultivariate context.

Review of Sociology 9 (2003)

56 MÁRTON MEDGYESI–PÉTER RÓBERT

Table 2. Ordered logit estimation of satisfaction with work

Model 1. Model 2.

Coefficients Std. error Coefficients Std.error

Women 0.299 0.046 0.206 0.065Age 0.001 0.013 -0.035 0.016Age squared 0.000 0.000 0.000 0.000Years of schooling -0.096 0.030 -0.116 0.034Years of schooling squared 0.003 0.001 0.004 0.0012. Income tercile (ref. cat. 1st tercile) 0.125 0.054 -0.134 0.0793. Income tercile (ref. cat. 1st tercile) 0.299 0.062 -0.018 0.090Supervisor 0.251 0.049 0.162 0.064Self- employed 0.306 0.084 0.273 0.109Part-time worker 0.081 0.069 -0.017 0.089

Trade union membership 0.073 0.043 0.086 0.060

Discrepancy: Job security -0.389 0.047 -0.349 0.068Discrepancy: High pay -0.416 0.049 -0.417 0.065Discrepancy: Advancement -0.297 0.046 -0.317 0.065Discrepancy: Interesting work -1.315 0.063 -1.598 0.095Discrepancy: Independence -0.314 0.062 -0.607 0.100Hard physical work -0.128 0.045 -0.116 0.063Stressful work -0.346 0.046 -0.436 0.064Good personal relations 1.212 0.046 1.206 0.064Transition societies -0.141 0.046 -3.077 0.685Interactions

Transitional – age 0.093 0.026Transitional – age squared -0.001 0.000Transitional – 2. Income Tercile 0.485 0.110Transitional – 3. Income Tercile 0.625 0.125Transitional – Supervisor 0.231 0.101Transitional – Discr. Interesting work 0.487 0.123Transitional – Discr. Independence 0.451 0.127_cut-point 1 -5.644 0.326 -6.948 0.414

_cut-point 2 -4.666 0.316 -5.965 0.405

_cut-point 3 -3.445 0.311 -4.735 0.401

_cut-point 4 -1.804 0.309 -3.078 0.399

_cut-point 5 0.468 0.308 -0.782 0.398

_cut-point 6 2.047 0.309 0.805 0.398

N 8110 8110

Chi2(23) 2753.26 2863.46

Prob > chi2 0.000 0.000

Pseudo R2 0.111 0.115

Note: coefficients significantly different from zero (at 0.05 significance level) are marked bold. Onlysignificant interactions are shown.

The subjective determinants, however, turn out to be very influential. All of the

five discrepancy measures affect satisfaction with work negatively. Thus, as expected

and as mentioned in the descriptive part of the paper, if respondents feel that the job

does not meet their preferences they will be less satisfied with work. The impact of this

attitude is present even if it is controlled for several status variables. The strongest

effect in this respect we have found for the case of not having an interesting job while it

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 57

is an important requirement of a good job according to the respondent. In addition,

people are less satisfied with work if it is physically hard and/or mentally stressful.

Good relations to fellows in work and/or to managers and supervisors, however,

increase satisfaction with work. These results are in line with the hypotheses as well.

The multivariate analysis confirms the results of the descriptive analysis;respondents in the former socialist countries are less satisfied with work in comparisonto those who live and work in more traditional market economies. Model 2 gives moredetails about these differences between the ‘old and new democracies’. One of thecharacteristic dissimilarities relates to the role of aging in the two types of societies. InModel 2 the age effects turn out to be significant and indicate basically that satisfactionwith work decreases as respondents are getting older (negative coefficient for the maineffect). But this relationship is not linear either; it decreases and turns back after a certainage (positive effect for the quadratic term of age). Thus middle-aged people seem to bethe less satisfied with work. However, this pattern is reversed for the respondents fromthe former socialist countries. Satisfaction with work increases basically with age in thetransition societies where gerontocracy was stronger and aging played a more effectiverole in status attainment. Although this effect is not linear either, the different pattern forthe two types of societies is obvious. Another characteristic difference between the twogroups of countries is that the income effects are significantly stronger in the transitionsocieties. This confirms our hypothesis that satisfaction with work is much more relatedto materialistic grounds in these countries. The third important difference is connected tothe position in the organization. Being a manager or a supervisor has stronger positiveimpact on satisfaction with work in the former socialist countries where these positionsinvolve probably more structural advantages, even in addition e.g. to higher earningsbecause the model controls for them. Thus, the multivariate analysis confirms thedescriptive findings for these predictor variables. Finally, the last distinction againunderlines the materialistic character of satisfaction with work in the post-socialistcountries where discrepancy for interesting and/or independent work decreasessatisfaction with work significantly less (as the negative coefficients for the main effectsand the positive coefficients for the interaction terms reveal). Even if respondents in thetransition societies feel their job to be less interesting or less independent than it wouldbe wishful, they will not be so much dissatisfied with work only for this reason.

So far, we could learn about marked differences in the mechanisms of generatingsatisfaction with work between transition societies and more traditional marketeconomies. In the next step, we have investigated whether this difference can be derivedfrom compositional effects, typical characteristics of the respondents that differ in thetwo clusters, or not.6 Thus we practically learn about the deviation in satisfaction withwork for two ‘imaginary’ respondents in the two country clusters but with the samesocial and demographic characteristics as well as with the same values and opinions. The

Review of Sociology 9 (2003)

58 MÁRTON MEDGYESI–PÉTER RÓBERT

6 For this purpose the predicted values of Model 2 were calculated for the two groups of countries. Whencalculating these predicted values, for the traditional market economies the value of the dummy variablewas set at 0; the values of the interaction terms were set at 0; for the transition societies the value of thedummy variable was set at 1; the values of the non-significant interaction terms were set at 0; the valuesof the significant interaction terms were set at their means. The values of all other variables in the modelwere set at their means.

results are shown in Figure 3. The distributions reveal that the observed difference islarger for the two groups of countries than the predicted difference. E.g. the proportion ofthose who were completely or very satisfied with work in the traditional marketeconomies was larger by 14 percent as compared to the proportion of the same categoryin the transition societies. The distribution based on the predicted values of Model 2,however, indicate a difference of 4 percent only. Thus we can conclude that a large partof the variation in the satisfaction with work is an outcome of compositional effects,namely that the major characteristics of the respondents in the country groups aredifferent. But still there are remaining ‘system specific’ dissimilarities as well.

Figure 3. Predicted probabilities for transitional countries and market-economies

Note: predicted probabilities were calculated holding explanatory variables at their means.

Model 3 repeats the analysis of Model 1 with investigating the differences amongthe country clusters in the more detailed manner. The main effects of the independentvariables do not differ from those discussed above for Model 1. As the estimates for thecountry group dummies indicate, respondents in the ‘Scandinavian fraction’ of thetraditional market economies are significantly more satisfied with work as comparedto those who live and work in the Western market economies (the reference category).But other groups of countries do not seem to differ according to the model.

This picture changes in Model 4 when the interaction terms are also included. Herethe main effects indicate that respondents in the transition societies and especiallythose in the peripheral post-socialist countries are significantly less satisfied with thework. The interaction terms reveal the mechanisms behind this finding. First of all,gender differences are more marked for Bulgaria and Russia where women aresignificantly more satisfied with work. Education seems to influence satisfaction withwork in a different way in these countries. Unlike the general pattern in Model 1 orModel 3 where satisfaction with work decreases basically as the level of schooling

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 59

increases, higher education makes people more satisfied with work in the peripheraltransition societies – though not linearly. There should be some (relatively) higher or(relatively) increased returns to human capital investments in these countries. But wecannot simply say that this is the higher earnings because income also increasessatisfaction with work in Bulgaria and Russia significantly and the estimates related toschooling and earnings are controlled for each other. In addition, supervisor positionmay further increase advantages in this group of countries, partly independently fromeducation and income. Finally, discrepancy with respect to interesting work decreasessatisfaction with work less in the peripheral post-socialist countries.

The group of the more developed transition societies also deviates in some respects. The

age pattern described above for Model 2 seems to be more characteristic for this cluster ofcountries, while this age pattern is not present for the peripheral transition societies – perhapsdue to the controls for earnings and supervisor position that may be strongly age related. Infact, income is a stronger predictor of satisfaction with work in the more developed transitionsocieties, too, but the estimate is smaller, on the one hand, and is not significant for those whohave the highest earnings in the 3rd income tercile, on the other. Thus, the relationshipbetween seniority and satisfaction with work seems to be different in the two groups offormer socialist countries. This relationship is apparently not linear in the more developedtransition societies. Finally, discrepancy with respect to independent work decreasessatisfaction with work less in the developed post-socialist countries.

Model 4 reveals much less deviations for the clusters of the traditional marketeconomies. Determinants of satisfaction with work differ very little for the peripheral marketeconomies. It seems that good relations to fellows at work and to managers and supervisorsmatter for them less. For the group of the three Nordic societies discrepancy with respect toadvancement at the work place decreases satisfaction with work less. Physically hard work,however, is a stronger negative predictor of satisfaction with work in these countries.

As a last step of the analysis, we have investigated again whether the dissimilarities insatisfaction with work can be derived from compositional effects, typically differingcharacteristics of the respondents in the five clusters of countries. The distributions basedonly on the predicted values from Model 4 are presented in Figure 4. We can observe thatpractically there is only a small deviation between the two groups of the developed marketeconomies; satisfaction with work is just a bit higher in the Nordic countries. Thus most ofthe observed differences (mentioned in the descriptive section of the paper) were due tocompositional effects. But if controlling for this compositional variation, respondents in theperipheral market economies are noticeably less satisfied with work. With respect to thetransition societies, observed frequencies in the previous figures above suggested thatsatisfaction with work is the lowest for the respondents in the less developed, peripheralfraction of these countries. Now we can see that respondents in the developed transitionsocieties are less satisfied with work than those living and working in the other post-socialistgroup of countries. In fact, satisfaction with work is the lowest in the developed transitionsocieties. This is a confirmation of the reference group theory arguing that people indeveloped post-socialist countries measure their circumstances against the traditional marketeconomies and compare themselves, accordingly, upward. Consequently, a relatively bettersituation does not necessary lead to a higher level of satisfaction for them. We will furtherelaborate on this finding in the discussion section of the paper.

Review of Sociology 9 (2003)

60 MÁRTON MEDGYESI–PÉTER RÓBERT

Table 3. Ordered logit estimation of satisfaction with work

Model 3. Model 4.

Coefficients Std. error Coefficients Std.error

Women 0.300 0.046 0.202 0.100Age -0.011 0.012 -0.032 0.026

Age squared 0.000 0.000 0.000 0.000

Years of schooling -0.092 0.029 -0.111 0.085Years of schooling squared 0.003 0.001 0.003 0.0032. Income tercile (ref. cat. 1st tercile) 0.145 0.055 -0.153 0.1223. Income tercile (ref. cat. 1st tercile) 0.324 0.063 -0.039 0.141Supervisor 0.269 0.049 0.148 0.096Self- employed 0.273 0.081 0.270 0.181Part-time worker 0.073 0.068 0.003 0.135

Trade union membership 0.017 0.047 -0.105 0.101

Discrepancy: Job security -0.332 0.047 -0.348 0.101Discrepancy: High pay -0.411 0.049 -0.340 0.100Discrepancy: Advancement -0.246 0.046 -0.458 0.099Discrepancy: Interesting work -1.311 0.063 -1.478 0.140Discrepancy: Independence -0.382 0.063 -0.491 0.140Hard physical work -0.147 0.045 0.020 0.096Stressful work -0.353 0.046 -0.414 0.098Good personal relations 1.161 0.046 1.174 0.099Country-groups (reference category= Developed Western countries)

Less developed transition societies 0.085 0.070 -3.998 1.230Developed transition societies -0.073 0.065 -2.381 1.143Peripheral market economies 0.074 0.085 -0.418 1.054

Developed (central) Nordic countries 0.259 0.064 1.103 0.988Interactions:

Less dev. tr.- women 0.495 0.148Less dev. tr.- years of schooling 0.315 0.139Less dev. tr.- years of school. sqrd. -0.011 0.005Less dev. tr.- 2. income tercile 0.811 0.175Less dev. tr.- 3. income tercile 1.215 0.193Less dev. tr.- supervisor 0.480 0.175Less dev. tr.- Disc. Interest. Wk. 0.564 0.188Developed tr.- age 0.122 0.039Developed tr.- age squared -0.002 0.000Developed tr.- 2. income tercile 0.357 0.169Developed tr.- Disc. Independence 0.452 0.186Peripheral market - Good relations -0.421 0.164Nordic countries-Disc. Advancement 0.427 0.139Nordic countries-Hard physical work -0.310 0.132_cut-point 1 -5.673 -6.632

_cut-point 2 -4.698 -5.647

_cut-point 3 -3.522 -4.454

_cut-point 4 -1.916 -2.818

_cut-point 5 0.245 -0.615

_cut-point 6 1.830 0.993

N 8110 8110

Chi2(23) 2647.74 2881.72

Prob > chi2 0.000 0.000

Pseudo R2 0.106 0.115

Note: coefficients significantly different from zero (at 0.05 significance level) are marked bold. Onlysignificant interactions are shown.

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 61

Figure 4. Predicted probabilities for country groups

Note: predicted probabilities were calculated holding explanatory variables at their means.

DISCUSSION

This study intended to provide a picture about the international variation in

satisfaction with work. It also intended to describe the various mechanisms, which

influence satisfaction with work in the different types of societies. When searching for

the explanatory factors that may have an impact on our dependent variable, both

objective and subjective determinants were considered such as social and demographic

characteristics of the respondents as well as their opinions and values related to a good

job. While we focused on international differences, we had an analytical design by not

considering any individual countries separately but by grouping these countries into

clusters based on theoretical and empirical grounds and by comparing these clusters of

nations. We hoped that our results testing several hypotheses based on a wide range of

literature would be more robust and substantial in this way. Even if we are convinced

about our research design in principle: it is not good to analyze all the countries one by

one, or it is not the best solution to cluster them by the dependent variable; we are aware

that the country clusters as developed here can be questioned in practice. A re-test of our

results with other kind of groups of countries, perhaps including also non-European

countries for having an even larger variation, is beyond the scope of this paper.

Results support most of our assumptions as well as they confirm the previous

findings of other analyses. Women are more satisfied with work than men. Those with

higher earnings and those being in managerial or supervisor position in the work

organization or being self-employed are more satisfied with work – in line with one of

the hypotheses arguing that higher social status results in higher satisfaction with

Review of Sociology 9 (2003)

62 MÁRTON MEDGYESI–PÉTER RÓBERT

work. This does not hold true for higher levels of schooling; the least and the most

educated are more satisfied with work. This supports the alternate hypothesis about the

role of reference group theory in determining satisfaction with work. Those with a

medium level of education are less satisfied with work than those with lower level of

education because they may have a higher horizon for reference and compare

themselves to those who have an even higher level of education. Discrepancy between

expectations for a good job and real features of the job decreases satisfaction with

work. The same holds true for the jobs that are physically hard or mentally stressful.

Good personal relations, however, increase satisfaction with work.

In the comparative perspective, respondents in the former socialist countries are

less satisfied with work. The mechanism behind this finding is the stronger materialist

value orientation under conditions when the GDP per capita is lower and the income

inequalities are higher in these countries. Consequently, certain status characteristics

like higher position in the work organizations or higher level of salaries have stronger

impact on satisfaction with work. Discrepancy with regard to post-materialist features

of the job decreases less the satisfaction with work as well. This especially holds true

for the less developed (peripheral) transition societies.

The multivariate analysis also revealed that a large part of the observed dissimilaritiesin satisfaction with work is an outcome of compositional differences in the workingpopulation of the country groups. When controlling for these differences, our resultsindicated a marked deviation in the mechanisms generating job satisfaction in the marketeconomies and in the new democracies. According to our descriptive and bivariatefindings, satisfaction with work was not necessarily the lowest of the market economies inthe peripheral countries, in Portugal or Spain. (See Figure 1, Figure 2 in the text or Figure

A1 in the Appendix.) But when applying a multivariate approach and filtering out thecompositional differences, satisfaction with work turned out to be lower in the peripheralmarket economies as compared to the more developed Western or Nordic countries. Thisfinding reveals a mechanism of generating job satisfaction, which is in line with thehypothesis based on the hierarchical impact of the level of economic development.

Our results are strikingly different for the former socialist countries. In the case ofthe two clusters of these countries, the descriptive and bivariate results do not makepossible to draw a definite conclusion whether satisfaction with work is higher in themore developed or in the peripheral group of countries. (See again Figure 1, Figure 2

in the text or Figure A1 in the Appendix.) However, when applying multivariateanalysis and filtering out the compositional differences, satisfaction with work turnedout to be lower in the more developed (central) nations. This result indicates amechanism generating job satisfaction, which is in line with the alternative hypothesisbased on the reference category theory. Even if economic conditions and workattributes are better in the more developed post-socialist countries, if we control forthis obvious advantage, respondents are relatively less satisfied with their job. Thisoccurs because – as we propose to interpret the result – people in the Czech Republic,Hungary, Poland, Slovenia have a reference group above them, they comparethemselves to the employees in the EU member countries. This is much less the casefor the respondents of Bulgaria and Russia where people have to be aware that acomparison to the Western market economies and any expectations on this ground are

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 63

much less realistic. The same view of satisfaction with work controlling formultivariate determinants and filtering out the compositional effects, as displayed onFigure 4, shows that satisfaction is higher even in the peripheral market economies.This increases the relevance to consider even the less developed market economies asreference cases for people in the EU applicant countries, in the cluster with lowestsatisfaction with work.

Obviously, alternative interpretation of our results could be offered. A usual list of

the so-called ‘unmeasured heterogeneities’ could also be produced on further

determinants and correlates of job satisfaction, we did not consider in this analysis. We

leave this to the readers of the article. As in many cases, it is also possible that other

methodology would have led to (partly) different results. We also intend to return to the

same data for additional analyses but this possibility is open for other researchers as well

because the data-file is available from the authors without any restrictions. One of the

future tasks is separating the objective and subjective predictors of satisfaction with

work in order to investigate the distinct role of these two groups of determinants in the

mechanism for generating job satisfaction. We are especially interested in finding out if

the objective or the subjective measures are more responsible for our result for the ‘new’

democracies in line with the reference group theory. At this point however, until new

evidences are not available, we are confident that applying the clustering design of

nations by political economy criteria like ‘old’ and ‘new’ democracies as well as by

political and economic geography like center and periphery makes sense and was a

useful decision. We were able to show cross-cutting effects of economic development

and of reference group theory on satisfaction with work and this picture may be at least

as conclusive as any other based on a comparison of each separate country.

REFERENCES

Beck, E. M.–Colclough, G. S. (1988): Schooling and Capitalism: The Effect of Urban EconomicStructure on the Value of Education. In Farkas, G.–England, P. (eds.): Industries, Firms, and

Jobs. Sociological ad Economic Approaches. New York: Plenum Press.Becker, G. S. (1975): Human Capital. A Theoretical and Empirical Analysis with Special

Reference to Education. Chicago: University of Chicago Press.Blanchflower, D. G.–Oswald, A. J. (1999): Well-Being, Insecurity and the Decline of American

Job Satisfaction. Paper presented at Cornell University, April.http://www.dartmouth.edu/~blnchflr/papers/JobSat.pdf

Borjas, G. J. (1979): Job Satisfaction, Wages and Unions. Journal of Human Resources, 14:21–40.

Clark, A. E. (1997): Job satisfaction and gender: Why are women so happy at work? Labor

Economics, 4: 341–372.Clark, A. E.–Oswald, A. J.–and Warr, P. B. (1995): Is Job Satisfaction U-Shaped in Age?

Journal of Occupational and Organizational Psychology, 69: 57–81.CSO (1999): Yearbook of International Statistics. Budapest: Hungarian Central Statistical Office.Curtice, J. (1993): Satisfying Work – If You Can Get It. In Jowell, R.–Brook, L.–Dowds, L.

(eds): International Social Attitudes: The 10th BSA Report. London: SCPR.

Review of Sociology 9 (2003)

64 MÁRTON MEDGYESI–PÉTER RÓBERT

Drobnic, S. (1997): Part-time work in Central and Eastern European Countries. In Blossfeld,H-P.–Hakim, C. (eds): Between Equalization and Marginalization. Women Working

Part-time in Europe and the United States of America, Oxford: Oxford University Press.Ester, P.–Halman, L.–Rukavishnikov, V. (1994): The Western world values pattern viewed

cross-nationally: A comparison of findings of the European and North American value studywith recent Russian data. WORC Paper 94.11.055/6. Tilburg: University of Tilburg.

Farkas, G.–England, P.–Barton, M. (1988): Structural Effects of Wages: Sociological andEconomic Views. In Farkas, G.–England, P. (eds.): Industries, Firms, and Jobs.

Sociological ad Economic Approaches. New York: Plenum Press.Freeman, R. B. (1978): Job Satisfaction As An Economic Variable. American Economic

Review, 68: 135–141Goldthorpe, J. (1982): On the Service Class, its Formation, and the Future. In Giddens,

A.–MacKenzie, G. (eds.): Social Class and the Division of Labour. Cambridge: CambridgeUniversity Press.

Greene, W. H. (1993): Econometric Analysis. Englewood Cliffs, N. J.: Prentice Hall.Hakim, C. (1991): Grateful slaves and self-made women: fact and fantasy in women’s work

orientation. European Sociological Review, 7 (2).Hakim, C. (1998): Social Change and Innovation in the Labour Market. Oxford: Oxford

University Press.Hradil, S. (1994): Soziale Schichtung und Arbeitssituation. In Geißler, Rainer (ed.): Soziale

Schichtung und Lebenschancen in Deutschland. Stuttgart: Verdinand Enke Verlag.Inglehart, R. (1990): Culture Shift in Advanced Industrial Society. New Yersey: Princeton

University Press.Kalleberg, A. L. (1977): Work Values and Job Rewards: A Theory of Job Satisfaction. American

Sociological Review, 42: 124–43.Kalleberg, Arne L.–Loscocco, Karyn A. (1983): Aging, Values and Rewards: Explaining Age

Differences in Job Satisfaction. American Sociological Review, 48: 78–90.Kalleberg, A. L.–Reve, T. (1992): Contracts and Commitment: Economic and Sociological

Perspectives on Employment Relations. Human Relations, 45 (9).Kertesi, G.–Köllõ, J. (2002): Economic Transformation and the Revaluation of Human Capital –

Hungary, 1986–1999. In De Grip, A.–Van Loo, J.–Mayhew, K. (eds.): The Economics of

Skills Obsolescence: Theoretical Innovations and Empirical Applications. Research LaborEconomics Vol. 21., Amsterdam: JAI Press.

Kolosi T.–Sági, M. (1998): Top Entrepreneurs and Their Social Environment. Acta

Oeconomica, 49 (3-4): 335–364.Long, J. S. (1997): Regression Models for Categorical and Limited Dependent Variables.

Thousand Oaks: Sage Publications.Lopreato, J.–Hazelrigg, L. (1972): Class, Conflict and Mobility. San Francisco: Chandler.Medgyesi, M.–Róbert P. (1998): Attitudes toward work: Longitudinal and international comparisons.

In Kolosi, T.–Tóth, I. Gy.–Vukovich, Gy. (eds.): Social Report 1998. Budapest: TÁRKI.Merton, R. K. (1968): Social Theory and Social Structure. 3rd Edition. Glencoe: Free Press.OECD (1999): Trends in Income Distribution and Poverty in the OECD Area. Paris: OECD.Polachek, S. W. (1981): Occupational Self-Selection: A Human Capital Approach to Sex

Differences in Occupational Structure. Review of Economics and Statistics, 63: 60–69.Róbert, P. (1998): Transformation processes in Eastern Europe: Social Stratification. In

Nieuwbeerta, P.–Ganzeboom, H.B.G. (eds.): Conference Transformation Processes in

Eastern Europe. Part 2. Social Stratification, The Hague: NWO/ESR.Róbert, P. (1999): Gritting the Teeth – Embittered with the Change of System. In Spéder, Zs.

(ed.): Hungary in Flux. Society, Politics and Transformation. Hamburg: Krämer Verlag.

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN PERSPECTIVE 65

Róbert, P.–Medgyesi, M. (1999): Changing attitudes, expectations and satisfaction in workrelations: comparison between 1989 and 1997, Hungary. Paper prepared for the conference‘Plant closures and downsizing in Europe’, Leuven, 28–30 January.

Simkus, A. (1981): Comparative Stratification and Mobility. International Journal of

Comparative Sociology, 22 (3–4): 213–236.Schömann, K.–Rogowski R.–Kruppe T. (1998): Labour Market Efficiency in the European

Union. Employment protection and fixed-term contracts. London: Routledge.Vecernik, J. (1999): Inequalities in earnings, incomes, and household wealth. In Vecernik,

J.–Mateju, P (eds.): Ten Years of Rebuilding Capitalism: Czech Society after 1989. Prague:Academia.

Vecernik, J. (2000): Work values and perceived conditions in CEE and EU countries. Paper

prepared for the ‘Household, Work and Flexibility’ project, Vienna, 7–9 May.World Bank (2000): Making Transition Work for Everyone: Poverty and Inequality in Europe

and Central Asia. Washington D.C.: World Bank.

APPENDIX

Figure A1. Work satisfaction in European countries covered by ISSP Work OrientationModule 1997

Notes: CH – Switzerland; DK – Denmark; E – Spain; N – Norway; S – Sweden; PL – Poland; D-W –West-Germany; GB – Great Britain; CZ – Czech Republic; I – Italy; BG – Bulgaria; F – France; P –Portugal; SLO – Slovenia; H – Hungary; RUS – Russia.The category ‘dissatisfied’ has been constructed by combining lower three categories (completelydissatisfied, very dissatisfied, fairly dissatisfied) of the original seven- point scale. Upper two categories ofthe original scale (completely satisfied, very satisfied) were also combined.

Review of Sociology 9 (2003)

66 MÁRTON MEDGYESI–PÉTER RÓBERT

Table A1. Orderings of countries with respect to macro- level variables

GDP per capita(USD, PPP, 1996)

Gini coefficient of households’net income (middle ’90s)

Percentage of trade unionmembers

Norway 25500 Denmark 0.22 Denmark 83.3%

Switzerland 24900 Sweden 0.23 Sweden 82.4%

Denmark 23100 Czech Republic 0.25 Norway 57.3%

Germany 21300 Norway 0.26 Russia 56.0%

France 20400 Switzerland 0.27 Slovenia 51.9%

Italy 20200 France 0.28 Bulgaria 39.3%

Sweden 19900 Germany 0.28 Czech Republic 36.3%

Great-Britain 19500 Hungary 0.28 Great-Britain 30.8%

Spain 15300 Slovenia 0.30 Italy 26.2%

Portugal 13800 Spain 0.30 Germany 24.5%

Slovenia 13200 Great-Britain 0.32 Hungary 20.7%

Czech Republic 12700 Poland 0.33 Switzerland 19.3%

Hungary 9300 Italy 0.35 Portugal 18.0%

Poland 6900 Portugal 0.36 France 17.5%

Russia 6700 Bulgaria 0.41 Poland 16.1%

Bulgaria 5000 Russia 0.47 Spain 15.3%

Sources: GDP per capita: CSO, 1999.Gini: Trends in Income Distribution and Poverty in the OECD Area. Paris, OECD 1999, except fortransitional countries and Portugal, where: Making Transition Work for Everyone. Washington, D.C., WorldBank, 2000.Percentage of trade union members: own calculation based on the ISSP 1997 data.

Review of Sociology 9 (2003)

SATISFACTION WITH WORK IN A EUROPEAN 67

Table A2. Distribution of the explanatory variables (% or means)

Less Developed EU Western- Nordic Total

developed transitional periphery Europe countries

transitional countries

countries

Objective variables (%)

Women 46.5 46.1 40.0 44.1 47.7 44.9

Supervisor 18.8 23.9 23.6 42.6 34.1 28.5

Self-employed 10.8 16.2 28.4 15.7 9.2 14.8

Part-time work 13.5 8.3 14.2 19.8 17.7 14.7

Union membership 47.4 31.0 16.7 23.7 74.5 38.5

Subjective variables (%)

Job security…..

….important 92.3 94.4 96.6 95.2 88.6 93.4

….holds true for the job 46.3 56.5 61.4 58.3 69.5 58.4

High pay….

….important 97.8 91.7 86.2 75.1 67.0 83.6

….holds true for the job 16.1 15.6 15.7 20.6 26.7 19.0

Good advancement…..

….important 64.3 67.7 85.4 73.1 48.2 67.8

….holds true for the job 14.1 15.7 21.3 19.0 19.3 17.9

Interesting Work…..

….important 80.2 91.0 89.6 97.3 97.1 91.0

….holds true for the job 52.1 65.6 72.1 77.3 79.6 69.4

Independence…..

….important 65.6 77.3 75.9 78.1 88.1 77.0

….holds true for the job 50.9 67.6 65.5 71.6 88.3 68.9

Find work physically hard 53.9 46.0 42.3 43.4 37.5 44.6

Find work stressful 32.6 33.4 36.8 36.9 37.4 35.5

Good personal relationsat the workplace 51.4 53.7 66.5 66.3 65.1 60.6

Continuous variables (means)

Age (years) 39.6 38.2 39.7 39.8 41.4 39.7

Years of schooling 11.7 12.1 9.8 11.9 12.3 11.6

Review of Sociology 9 (2003)

68 MÁRTON MEDGYESI–PÉTER RÓBERT