Content uploaded by Ramón J. Torregrosa
Author content
All content in this area was uploaded by Ramón J. Torregrosa on Feb 06, 2020
Content may be subject to copyright.
1
Re-evaluating Job Satisfaction
Cristina Pita
Ramón J. Torregrosa
Universidad de Salamanca
January 2020
Abstract
Empirical research on the determinants of job satisfaction has focused on workers and job
characteristics as well as on labor market institutional factors. To our knowledge, all these
studies are based on cross-sectional or panel data drawn from surveys in which workers are
asked to report the level of satisfaction with their jobs in a categorical scale. In this article we
reevaluate previous findings using a new methodology specifically designed to measure the
relative performance of groups of individuals whose levels of satisfaction are described by a set
of ordered categories. In particular, we apply the technique of comparison of categorical data,
the Balanced Worth, posed by Herrero and Villar (2018). Our data set consists of a large sample
of workers from the sixth wave of the European Working Conditions Survey, which covers 35
countries. Many of our results confirm previous findings in the empirical literature on job
satisfaction but we also intend to shed new light upon some controversial issues, such as the
interactions between the worker’s gender, education, age, the type of employment contract,
and job satisfaction.
JEL Classification: I31, J53
Key words: job satisfaction, worker and job characteristics, Balanced Worth.
2
1 Introduction
Job satisfaction has drawn increasing attention over the last decades because of its expected
effects on job performance and worker´s well-being. Understanding the factors involved in job
satisfaction becomes crucial to improving well-being given that most people spend a large part
of their lives at work. In addition, job satisfaction could have positive implications on workers
behavior in firms and labor organizations. Employees who are more satisfied with their jobs
should perform better (George and Jones, 2008) and face a lower probability of quitting (Carsten
and Spector, 1987) than those who feel less satisfied.
Nevertheless, whereas sociologists and industrial psychologists have traditionally done
extensive research on job satisfaction and workers wellbeing, for a long time, economists did
not address this topic because studying job satisfaction involved taking into account both
objective and subjective factors. Regarding “the answers that people give when asked questions
about how happy they fell with life or how satisfied they feel with their job and work”, Oswald
(1997) remarks that “there are limitations to such statistics, but, if the aim is to learn about what
makes people tick, listening to what they say seems likely to be a natural first step. Sources of
information exist that have for many years recorded individuals' survey responses to questions
about subjective well-being. These responses have been studied intensively by psychologists,
studied a little by sociologists, and ignored by economists. Some economists may wish to defend
this neglect by emphasizing the unreliability of such data. Most, however, are probably unaware
that data of this sort are available and have not thought of how such empirical measures might
be used in their discipline” (Oswald, 1997).
Thus, reluctant to use data on subjective feelings (Freeman, 1978; Sloane and Williams, 2000),
economists relied the study of job satisfaction to other disciplines. However, following
Hamermesh´s seminal paper (Hamermesh, 1977), job satisfaction has received more attention
within the Economics and Business literature because of its expected implications increasing
workers productivity. Nowadays the motivation to analyze job satisfaction has been well
reported due to its effects not only on productivity, but also on absenteeism, quits, turnover,
and mobility (Freeman, 1978; Warr, 1999). In fact, Freeman (1978) suggested that job
satisfaction should be considered as an economic variable, albeit it would require a more
sophisticated and careful analysis than other standard variables.
Since then, a growing literature has intended to analyze differences in individuals' self-reported
satisfaction with their jobs, just as scholars have used data on subjective well-being to explore
the determinants of happiness and life satisfaction. Both lines of research are obviously closely
related and have grown side by side (Fernández Macías and Muñoz de Bustillo, 2014). Much of
this literature has been empirical and focused either on the determinants or the implications of
job satisfaction. The evidence so far has mainly been based on cross-sectional or pooled data
drawn from surveys made to workers. We can either find studies that analyze the workers
opinion over objective and measurable characteristics of jobs (Muñoz del Bustillo and Fernández
Macías, 2005) and another line of research based on the subjective evaluations of workers´
wellbeing.
Results have often been striking because they have not always followed the expected pattern,
very much in line with recent research on individual happiness and well-being, and seem to
suggest that job satisfaction could be considered a reasonable measure, but not a perfect
predictor, of job quality (Osterman, 2013). For example, we would expect to find better jobs and
better work environment in countries with higher levels of production and income. Indeed,
countries with higher GDP per capita should have workers with higher levels of job satisfaction.
However, as recent articles disclose, higher income levels are not perfectly correlated to higher
3
levels of either happiness or well-being (Blanchflower and Oswald, 2011). More surprisingly, as
income levels have risen over time in the USA and the UK, job satisfaction has not increased
accordingly (Oswald, 1997). Indeed, Clark (2005) found a downward trend on reported
satisfaction across European countries. Researchers have also found that women report a higher
level of job satisfaction than men although there is abundant empirical evidence proving that
they face worse conditions in terms of pay and other nonmonetary job characteristics than their
male colleagues. Clark (1997) attributes the gender differential in job satisfaction to the fact that
women expectations in the labor market are lower than those of their male counterparts. Just
to mention another counterintuitive result, more educated workers who hold better jobs have
not necessarily reported higher levels of job satisfaction (Clark and Oswald, 1996).
Our goal in this paper is to review previous research on job satisfaction using a new methodology
provided by Herrero and Villar (2018) specifically designed to compare distributions of
categorical data. We intend to shed some new light upon this topic analyzing self-reported job
satisfaction in a large sample of workers from 35 countries, which is the last available wave of
the European Workers Condition Survey (EWCS). In particular, we reevaluate the interaction
between worker and job characteristics and the level of job satisfaction that workers report.
Both workers’ personal characteristics, such as gender, marital status, age, education, and
health, and characteristics of the job itself, such as income, hours of work, union membership,
professional status, and activity sector have been considered explanatory variables in the job
satisfaction regressions. Among worker characteristics, we focus on those which have been
extensively studied in the past, which are gender, age, and education. In second place, we
consider basic characteristics of the job and labor contract such as whether the worker is an
employee or self-employed, the job is full-time or part-time, and the sector and duration of the
contract. In addition, we measure and rank the relative degree of job satisfaction across the 35
countries that were included in the last wave of the EWCS.
The paper proceeds as follows. In the next two sections, we describe the methodology and the
data set. The following section shows our results. First, we analyze cross national differences in
job satisfaction. Secondly, we focus on workers characteristics, such as gender, age, and
education. Finally, we analyze the relationship between the type of employment contract in
terms of workday hours or duration, as well as employment status and sector, and job
satisfaction. We also analyze the interaction among some of these variables, devoting close
attention to gender differentials. Section 5 summarizes our main findings.
2 The Balanced Worth procedure
Whereas previous research on job satisfaction has extensively analyzed its causes and
consequences, we focus on its measurement and on establishing rankings among different social
groups. The value added of this paper regarding the previously published articles written on this
topic consists of applying a new methodology to rank the job satisfaction of European workers
using a method specifically designed to compare distributions of categorical data.
Following Herrero and Villar (2018), let us consider the problem of comparing the distributions
of g different groups over a set of k categories. The following table shows the information we
can draw from our data set, where the first row represents categories, which are ordered form
best to worst, the first column corresponds to the different groups and is the frequency in
which category j appears in group i, where = 1, = 1,2, … , .
4
…
…
…
…
Our comparison problem consists in evaluating the relative dominance of these frequencies
among them. This is an extension of the concept of Net Difference introduced by Lieberson
(1976) for the case of two groups. Hence, let
= + +++++++()
be the probability that a randomly chosen individual from group i belongs to a better category
than a randomly chosen individual from group j. On the other hand, the probability that the
randomly chosen individuals from both groups belong to the same category is
=+++.
Where = and + + = 1. The procedure posed by Herrero and Villar (2018)
consists in choosing one individual from each group at random and comparing its category with
that of a randomly chosen individual from all the other groups. In the pairwise comparison, when
a randomly chosen individual of group i belongs to a higher category than a randomly chosen
individual of group j, we say that the distribution of group i dominates the distribution of group
j. If both individuals belong to the same category (with probability ), each group is declared
dominant with probability 0.5. Assuming that the cost of choosing at random one individual from
group h is denoted by , the expected cost of choosing at random an individual from group i is
not higher than the expected cost of choosing at random an individual of worse category from
the other groups if
+
++ +
+
++ +
.
Hence, denoting by the maximum value that we are willing to pay for choosing an individual
of a better category from group i than from the other groups, saturates the previous
inequality, that is +
= +
.
Extending this criterion to the other groups, substituting the cost by the willingness to pay,
, we can write the following linear equations system:
+
= +
+
= +
+
= +
Which is a homogeneous linear system that can be written as M v = 0, where
5
=
+
+
… +
+
+
…+
+
+
…
+
and =,, … , . As the vectors that conform M are linearly dependent, M is singular,
thus, the homogeneous system M v = 0 has a non-trivial solution. To amend the indetermination
of the solution we add an additional equation associated to the normalization of the marginal
willingness to pay vector, =
.0F
1 Herrero and Villar (2018) prove that the solution to this
system exists, is unique and is given by =
,
, … ,
such that
=
, and
= +
+
,= 1,2, … , .
We call the Balanced Worth Vector (BWV)2. Note that the component of the BWV of group i
is given by the ratio between the relative advantage of the distribution of group i with respect
to the other groups and the relative disadvantage of the distribution of group i with respect to
the other groups. The main contribution of the BWV is that, according to the probability of the
domination criterion, it ranks the different groups by means of a cardinal measure. In contrast
to the score methods previously used to compare categorical distributions, the BWV not only
ranks the group dominance but it also provides endogenous information about the intensity of
such dominance by means of the cardinality of its components.3 Furthermore, in this paper we
calculate the first differences of the ranked BWV components (∆BWV*). This measure allows us
to explain the significance that groups have in explaining a variable and, in the presence of a
large number of groups, the formation of clusters according to their level of homogeneity. For
instance, let us suppose a set of groups that exhibit similar distributions over the categories of
the variable. In that case we can assert that belonging to each of these groups does not exert
any influence over the variable, or, in other words, groups are not relevant in explaining the
variable. In such a case, the similar distributions make the BWV to be a close-to-ones vector
making the ∆BWV* (that has one dimension less) to be close to the vector of zeroes, and the
variability of the components of the BWV (measured by its coefficient of variation) to be low.
Otherwise, the higher the components of ∆BWV*, the more different the distribution among
groups, the higher the variability of the components of the BWV and the higher the likelihood
that such groups explain the variable. Moreover, the ∆BWV* will be useful in finding cluster
patterns, both of homogeneity and heterogeneity, when there exists a large number of groups.
Thus, the components of the ∆BWV* indicate whether or not the groups studied exert influence
over a given variable.
3 The data
The data set we use is the sixth wave of the European Working Conditions Survey (EWCS). The
EWCS is conducted by the European Foundation for the Improvement of Living and Working
Conditions (Eurofound). Since its first wave in 1990, the EWCS has provided a complete overview
1 This constraint also allows us to equalize to one the average willingness to pay
1= 1
.
2 Across the paper BWV* refers to a realignment of the BWV where its components are ranked for the
higher to the lower.
3 The BWV has other desirable properties, such as anonymity, symmetry and monotonicity (see Herrero
and Villar, 2013 and 2018).
6
of working conditions in Europe. The survey is based on a questionnaire which is administered
face-to-face to a random sample of 'persons in employment', both employees and self-
employed, which is representative of the working population in each EU country.
Composition of the countries included in the survey has changed over time, adding new
countries in recent waves that goes beyond the scope of the EU. In 2015 the survey was
conducted in 35 countries: EU28, Norway, Switzerland, Albania, the former Yugoslav Republic
of Macedonia, Montenegro, Serbia and Turkey.
The EWCS has 106 questions, some of them being closely related to job satisfaction. Among
these questions, Q88 stands out as the best predictor of overall job satisfaction: “On the whole,
are you very satisfied, satisfied, not very satisfied or not at all satisfied with working conditions
in your main paid job?”
Our main goal in this paper is to calculate the BWV of question Q88 for different social groups
determined by the country, the sector, the type of contract, the gender, the age, or the level of
education of the workers interviewed in the EWCS. The web site of the Instituto Valenciano de
Investigaciones Económicas (IVIE) provides a freely available algorithm to calculate the BWV
(http://www.ivie.es/balanced-worth/). In order to make reproducible all our calculations we
provide the main relative frequency matrices at the beginning of each of sections devoted to
the analysis.
Questions regarding job satisfaction are quite similar in different surveys, asking workers for an
overall evaluation of job satisfaction, and offering them several categorical responses. According
to Fernández-Macías and Muñoz de Bustillo (2014), “the single measure can be understood as
an overall evaluation of the job, with the workers themselves averaging the positive and
negative attributes according to their own preferences”.
Across countries and surveys, most workers report to be highly or quite satisfied with their jobs.
Freeman (1978) remarked that most studies found a relatively high level of job satisfaction in
countries (above 75%), that is, most workers are satisfied or very satisfied with their main jobs,
whereas only a minority of about 10 percent of workers reported dissatisfaction. In our data set,
43850 workers were interviewed. Their answers to question Q88 are the following: 24.79% of
workers report to be very satisfied, 58.41% report to be satisfied, 13.08% report to be not very
satisfied, 3.02% report to be not at all satisfied, and 0.6% report no opinion or refuse to answer
this question.
Before beginning with the analysis, we must make a caveat. Although we cannot pretend to be
measuring job quality (Osterman, 2013), we are addressing a topic which is extremely important
to individuals, that is, how they feel about their work. We focus on worker´s perceptions and
feelings that could not have been measured by anyone but the worker herself. However,
reported satisfaction could be subject to important biases depending on workers personality
and attitude towards life. Thus, it would be desirable to have a method to measure how biased
a worker’s perceptions could be. In this regard, our analysis would be enriched if anchorage
vignettes were provided and made available with this survey (Hinz et al., 2016).
4 The Analysis
In the following subsections, we analyze job satisfaction across countries, genders, ages, levels
of education, sectors, working times, employment status and types of labor contracts. We will
pay special attention to gender issues along the different sections of the paper.
7
4.1 Country Effects
In this section, we focus on the construction of the BWV for different countries in order to
measure and rank job satisfaction across countries. We construct the distributions for each
country from the data provided by the EWCS. The number of questionnaires delivered in each
country ranges from 1001 in Finland to 3364 in Spain. Even when we disregard responses like
“DK/no opinion” and “Refusal”, the number of categorical effective responses, “very satisfied”,
“satisfied”, “not very satisfied”, and “not at all satisfied”, remains high enough to consider the
samples statistically representative.4
The Eurofound Report on EWCS states that “differences between Member States on these
factors are substantial. Changes over time do not illustrate upward convergence on all
dimensions of job quality.” Previous studies have found that country differences remain even
after controlling for individual and job characteristics. Indeed, most studies reveal large country
fixed effects (Ahn and García, 2004). According to Pichler and Wallace (2009), there exists a
country premium and a country penalty. They find that the predicted level of job satisfaction
based on individual-level variables in each country is higher or lower than the observed values
of job satisfaction.
Roughly speaking, we could say that workers from Northern European countries show higher
levels of job satisfaction, followed by Continental European Countries (and the UK) and by
Mediterranean countries (including France), and finally by Eastern European countries. Our
results will be in line with those of previous studies when we reevaluate the relative
performance of workers from 35 countries using the BWV.
Figure 1 shows the answers to Question 88 in the 35 countries where the 2015 wave of the EWCS
was undertaken.5
Figure 1: Job satisfaction according to EWCS (Eurofound)
4 The percentage of categorical effective responses is 99.2% on average with a standard deviation of
0.82%.
5 The Eurofound webpage provides an interactive data visualization tool that can be easily used to make
graphs and figures with the survey data.
8
Table 1 shows the frequencies and the BWV evaluation for question Q88, for the 35 European
countries surveyed in the EWCS ordered from the highest to the lowest BW component.
Table 1: Distributions, BWV* and ∆BWV* of Job satisfaction for the 35 EWCS countries
The most striking conclusion from the results shown in Table 1 is the high level of variability of
the BWV across countries, with a range of 1.103 for the BWV and a coefficient of variation of
25.56%, which suggests that a worker´s “country” exerts significant effects on job satisfaction.
Moreover, the vector ∆BWV* in Table 1 discloses other interesting patterns in the ordination
induced by the BWV. The components of the ∆BWV* in Table 1 are shown in Figure 2, where
Ranking
Country
Ve ry sati sfie d Satisfied Not ve ry sati sfied
Not at all satisfied
BWV*
∆BWV*
1
Denma rk
0.4730 0.4340 0.0810 0.0120 1.5902 -
2
Norway
0.4372 0.4976 0.0555 0.0097 1.5564 0.0338
3
Austria
0.4102 0.5117 0.0645 0.0137 1.4527 0.1037
4
Ireland
0.4097 0.4962 0.0741 0.0200 1.4144 0.0383
5
Switzerland
0.3792 0.5020 0.0988 0.0200 1.2930 0.1214
6
UK
0.3718 0.5158 0.0908 0.0216 1.2877 0.0053
7
Netherlands
0.3148 0.6053 0.0595 0.0205 1.2236 0.0642
8
Czech Rep
0.3130 0.5910 0.0840 0.0120 1.1947 0.0288
9
Cyprus
0.3333 0.5220 0.1267 0.0180 1.1507 0.0440
10
Fi nla nd
0.2770 0.6400 0.0780 0.0050 1.1467 0.0040
11
Belgium
0.2778 0.6080 0.0872 0.0270 1.0899 0.0569
12
Malta
0.2931 0.5633 0.1186 0.0249 1.0733 0.0166
13
Germany
0.2755 0.5981 0.1106 0.0159 1.0704 0.0029
14
Luxem bourg
0.2920 0.5380 0.1280 0.0420 1.0256 0.0448
15
Sweden
0.2663 0.5836 0.1301 0.0200 1.0168 0.0088
16
Bulgaria
0.2474 0.5776 0.1496 0.0254 0.9483 0.0686
17
Poland
0.2026 0.6655 0.1072 0.0247 0.9344 0.0139
18
Estonia
0.1747 0.7196 0.0928 0.0130 0.9283 0.0061
19
Slovakia
0.2093 0.6309 0.1466 0.0131 0.9114 0.0169
20
Spain
0.2309 0.5815 0.1448 0.0429 0.9000 0.0114
21
Portugal
0.1750 0.6843 0.1202 0.0205 0.8809 0.0191
22
Hung ary
0.1805 0.6720 0.1174 0.0301 0.8782 0.0027
23
Slovenia
0.2022 0.6211 0.1411 0.0356 0.8724 0.0058
24
Croatia
0.2214 0.5701 0.1663 0.0421 0.8595 0.0129
25
France
0.2079 0.5862 0.1600 0.0459 0.8417 0.0179
26
Italy
0.1736 0.6484 0.1333 0.0447 0.8264 0.0153
27
Greece
0.2159 0.5542 0.1878 0.0422 0.8252 0.0012
28
Romania
0.1125 0.7675 0.1011 0.0189 0.8152 0.0100
29
Lithuania
0.1712 0.6356 0.1802 0.0130 0.8130 0.0022
30
Turk ey
0.1693 0.6284 0.1627 0.0396 0.7929 0.0201
31
Latvia
0.1440 0.6646 0.1708 0.0206 0.7757 0.0172
32
Montenegro
0.1751 0.5845 0.1952 0.0453 0.7561 0.0196
33
Serbia
0.1781 0.5333 0.2319 0.0568 0.7055 0.0507
34
Macedonia
0.1586 0.5362 0.2349 0.0704 0.6618 0.0436
35
Albania
0.1174 0.4393 0.3290 0.1143 0.4874 0.1745
Coef.var
25.56
9
the number in the horizontal axis represents the difference of the BWV components of the BWV-
ranked countries j and j - 1.
Figure 2: ∆BWV* components of job satisfaction for the 35 EWCS countries
Figure 2 shows that job satisfaction across the EWCS countries, measured by means of their
BWV, can be broken down into several clusters. The first fifteen ranked countries (from Denmark
to Sweden) combine high and low differences by adjacent pairs, suggesting a considerable level
of variability and heterogeneity in the perception of job satisfaction among them (for instance
the partial coefficient of variation6 for these countries is 18.84%). Another characteristic of this
cluster of countries is that all their BWV components are above one (the mean). A second group
of countries, from Bulgaria to Montenegro, have lower variability (the partial coefficient of
variation of the BWV in this case is 5.71%). Finally, the cluster formed by Serbia and Macedonia
shows moderate differences with the previous cluster and between them, while Albania exhibits
the worst performance in the ranking and a high difference with respect to Macedonia.
In addition, the calculation of the ∆BWV* allows us to single out another relevant feature in the
clusters of countries that we have encountered, that is, we can identify pairs of countries where
the perception of job satisfaction is similar. For instance, ∆BWV* is below 0.01 for Italy-Greece,
Romania-Lithuania, Portugal-Hungary, Malta-Germany, Cyprus-Finland, Switzerland-UK,
Hungary-Slovenia, Poland-Estonia and Luxembourg-Sweden, which means that in these pairs of
countries workers report approximate levels of job satisfaction.
Both the variability of the BWV for the different countries and the existence of several clusters
(two heterogeneous and one homogeneous) suggest that the “country factor” is relevant
explaining job satisfaction in the EWCS. Although previous studies use surveys undertaken in
different sets of countries, our results are in line with former research outcomes. Ahn and García
(2004) also rank Denmark and Austria as the top performers in terms of workers job satisfaction
but their analysis was done with data from 14 countries. Sousa-Poza and Souza-Poza (2000),
using data from the 1997 International Social Survey Program (ISSP), which covers 21 countries,
find that workers are quite satisfied in all countries, and Denmark has the highest level of job-
satisfaction, the US is ranked as 7th, Britain 15th, Japan 19th, and Russia 20th. They also find
6 By the partial coefficient of variation we refer to the coefficient of variation of the BWV components of
a subset of countries.
10
that job satisfaction has declined in Germany, Norway, and the US in the 1990s when they
compare results with the 1989 ISSP.
4.2 Job satisfaction and gender
Much has been written about the relationship between gender and job satisfaction. In general
terms, females experience significantly more job satisfaction than males (Clark, 1997; Sloane
and Williams, 2000). Men report to be less satisfied with their jobs although evidence supports
that they have better jobs than women. This unexpected finding has become known as the
gender/job-satisfaction paradox and it has had several plausible explanations in the job
satisfaction literature. Clark (1996) and Gaziougly and Tansel (2006) remark the differences in
the qualifications and the types of jobs of men and women. For cultural reasons, women who
are dissatisfied at work may find it easier than men to quit their jobs. Besides, men and women
could perceive their level of job satisfaction questions in a different way because their
expectations about their job could be different (Clark, 1996). Clark (1997) finds empirical
evidence suggesting that women have lower expectations and concludes that their lower
aspiration is the principal reason for higher job satisfaction among women in the UK. Sousa-Poza
and Sousa-Poza (2003) find that the gender differential in job satisfaction was higher in the past
when women expectations in the labor market were much lower than those of men whereas
nowadays there is a more subtle difference between males and females.
Does the gender/job satisfaction paradox hold when we calculate the BWV of males and females
using the EWCS dataset?
Table 2 shows the frequencies and the BWVs for men and women’s job satisfaction in the five
most recent waves of the EWCS. For each wave, we calculate the difference between the female
and the male components of the BWV in order to analyze the comparative evolution of job
satisfaction between genders.
Table 2: Job satisfaction by gender in the last 5 EWCS
The results in Table 2 provide a clear answer to the previous questions. Overall, we can say that
there exists a gender differential that decreases over time, just in line with the results of Sousa-
Poza and Sousa-Poza (2003).
2015
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
Men 0.2446 0.5895 0.1340 0.0318 0.9859
Women 0.2547 0.5868 0.1295 0.0291 1.0141
2010
∆BWV
0.0282
Men 0.2215 0.5671 0.1657 0.0458 0.9697
Women 0.2365 0.5737 0.1564 0.0334 1.0303
2005
∆BWV
0.0606
Men 0.2279 0.5599 0.1651 0.0471 0.9713
Women 0.2468 0.5572 0.1584 0.0376 1.0287
2000
∆BWV
0.0574
Men 0.2570 0.5491 0.1520 0.0419 0.9651
Women 0.2877 0.5312 0.1448 0.0362 1.0349
1995
∆BWV
0.0698
Men 0.3376 0.5316 0.1014 0.0293 0.9747
Women 0.3651 0.5066 0.0976 0.0307 1.0253
∆BWV
0.0507
11
4.3 Job satisfaction and age
The relationship between age and job satisfaction is also controversial: some studies show it is
a U-shaped relationship (Clark, 1996; Clark et al., 1996; Blanchflower and Oswald, 2004;
Ghinetti, 2007) whereas other researchers (Johnson and Johnson, 2000) reach the conclusion
that job satisfaction increases with age.
Is job satisfaction U-shaped in age? Our answer is “yes”. Table 4 shows the frequencies and the
BWV for the different groups of age. It is noticeable how the components of the BWV decrease
as workers become older reaching its bottom value at ages in the range of 46-55 and increasing
for older workers, which is consistent with previous findings not only in the job satisfaction
literature but also in research done on happiness and well-being.
Table 4: Job satisfaction by age and BWV
In Figure 3 we can observe the U-shaped relationship between job satisfaction and age.
Figure 3: BWV for the EWCS of different age intervals
Furthermore, we have analyzed the interaction between age, gender and job satisfaction. Tables
7 and 8 show the results of the BWV for different age groups of both men and women. For both
genders, we can still observe the U-shapedness with younger and older workers reporting higher
levels of job satisfaction than middle-aged ones. Regarding the dispersion of the BWV
components, the respective coefficients of variation reveal similar degrees of concentration
(9.78% for men, 7.04% for women and 7.99% for the whole sample). There are, however, slight
differences between genders.
Group
Ve ry sati sfie d
Satisfied
Not very sat isfie d Not at all satisfied BWV
< 25 0.2935 0.5491 0.1299 0.0275 1.0476
26 - 35 0.2580 0.5794 0.1332 0.0294 0.9787
36-45 0.2356 0.6020 0.1312 0.0311 0.9420
46-55 0.2400 0.5901 0.1354 0.0345 0.9380
56-65 0.2405 0.5988 0.1324 0.0283 0.9530
> 65 0.3132 0.5645 0.1020 0.0204 1.1408
12
Table 5: Job satisfaction by age and BWV for men
Table 6: Job satisfaction by age and BWV for women
Figure 4 illustrates the differences between both genders at different ages. Women report to be
happier at work at younger ages whereas men become happier when they grow older.
Figure 4: BWV for men and women of different age intervals
Furthermore, Table 7 provides the BWV for men and women in each age group. Women of
different ages report higher job satisfaction than their male counterparts except when workers
are over 65 years old. In both calculations, we find that men and women have more similar
feelings towards their jobs in terms of satisfaction when they are in the age range between 36
and 65, as Figure 4 shows, and they differ more when they are either younger or older. Young
and middle-aged females report higher levels of job satisfaction than young and middle-aged
males but this tendency is reversed for older workers. Over 65, men who continue working
report to be more satisfied with their jobs than women.
Men
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
< 25 0.2740 0.5605 0.1365 0.0291 1.0113
26 - 35 0.2459 0.5848 0.1383 0.0310 0.9587
36-45 0.2307 0.5979 0.1393 0.0320 0.9316
46-55 0.2381 0.5947 0.1295 0.0376 0.9472
56-65 0.2390 0.5995 0.1335 0.0280 0.9592
> 65 0.3350 0.5408 0.1029 0.0213 1.1921
Women
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
< 25 0.3148 0.5366 0.1227 0.0259 1.0901
26 - 35 0.2706 0.5737 0.1280 0.0277 1.0006
36-45 0.2400 0.6064 0.1234 0.0303 0.9517
46-55 0.2419 0.5856 0.1410 0.0315 0.9292
56-65 0.2421 0.5982 0.1311 0.0286 0.9474
> 65 0.2835 0.5983 0.0991 0.0191 1.0810
0.0000
0.2000
0.4000
0.6000
0.8000
1.0000
1.2000
1.4000
< 25 26 - 35 36-45 46-55 56-65 > 65
Q88/Men Q88/Women
13
Table 7: BWV by levels of age and gender
4.4 Job satisfaction and education
Previous research on the effects of education on job satisfaction has drawn unexpected and
controversial results. Whereas Clark (1996, 1997), Clark and Oswald (1996), Sloane and Williams
(2000) and Sousa-Poza and Sousa-Poza (2003) conclude that job satisfaction decreases on the
education level, Johnson and Johnson (2000), Vila and García-Mora (2005) and Fiorillo and
Nappo (2011) find a positive relationship between the two.
Although it seems difficult to find a plausible explanation for a negative relationship between
education and job satisfaction, some authors suggest that job satisfaction depends on up to
what point worker´s expectations of educated workers are fulfilled. Clark and Oswald (1996)
concluded that, “holding income constant, satisfaction is declining in the level of education” and
attributed this result to the higher aspirations that people with higher levels of education have.
Workers’ income always plays a crucial role in these analyses.
Using a sample of Italian workers, however, Fiorillo and Nappo (2011) find that bachelor’s
degree holders report higher levels of job satisfaction than individuals with lower education or
no education at all. Vila and García-Mora (2005) analyze a survey of Spanish workers to explore
the effects of workers’ education on satisfaction with different aspects of their jobs and find that
the effects of the education level on job satisfaction depend on the aspect of the job considered
and workers’ perceptions of the match between education and employment are relevant
determinants of job satisfaction.
Both Johnson and Johnson (2000) and García-Mainar and Montuenga (2015) go a step further
and focus on the effect of over-qualification on job satisfaction. Using data from the American
Postal Workers Union, Johnson and Johnson (2000) suggest that self-perceived over-
qualification has a negative effect on job satisfaction. However, García-Mainar and Montuenga
(2015)) find that over-qualified works do not feel less satisfied than those who are adequately
matched to their jobs.
In this scenario, we are glad to find a clearly increasing relationship between job satisfaction and
education. The following tables show these results.
Table 8: BWV for job satisfaction by education and gender for the whole sample
BWV < 25 26 - 35 36-45 46-55 56-65 > 65
Men 1.0000 0.9604 0.9789 0.9998 0.9962 1.0410
Women 1.0000 1.0396 1. 0211 1.0003 1.0039 0.9590
Dif 0.0000 -0.0791 - 0.0421 -0.0005 -0.0077 0.0820
Whole sample
Ve ry sati sfied
Satisfie d
Not very sat isfie d
Not at all satisfie d
BWV*
∆BWV*
Doctorate
0.4034 0.4952 0.0845 0.0169 1.3331
Mast e r
0.3466 0.5488 0.0869 0.0178 1.2020 0.1311
Bachelor or equivalent
0.3063 0.5717 0.1022 0.0197 1.0942 0.1078
Short-cycle tertiary education
0.2909 0.5762 0.1127 0.0202 1.0492 0.0449
Post-secondary non-tertiary
0.2519 0.5987 0.1241 0.0253 0.9576 0.0917
Upper secondary education
0.2223 0.6102 0.1379 0.0295 0.8864 0. 0711
Lower secondary education
0.2088 0.5758 0.1657 0.0497 0.8045 0.0819
Primary education
0.1534 0.5777 0.2036 0.0653 0.6730 0.1315
Coe f. Va r
21.48
14
Table 9: BWV for job satisfaction by education and gender for men
Table 10: BWV for job satisfaction by education and gender for women
Tables 10, 11 and 12 show both the BWV* and ∆BWV* for job satisfaction by levels of education
for the whole sample and for each gender, respectively. In all tables groups are ranked
identically, associating higher job satisfaction with higher levels of education. We also find a
higher degree of variability in the BWV components for males than for females.
Figure 5 shows the ∆BWV for different levels of education. The number in the horizontal axis
denotes the difference between doctorate and master (1), master and bachelor or equivalent
(2), bachelor or equivalent and short-cycle tertiary education (3), short-cycle tertiary education
and post-secondary non tertiary education (4), post-secondary non tertiary education and upper
secondary education (5), upper secondary education and lower secondary education (6) and
lower secondary education and primary education (7).
For the whole sample, the largest increases in job satisfaction are associated with having a
doctoral degree versus a master´s degree or, on the other extreme, having accomplished
secondary versus primary education. However, we detect large differences in male and female
workers underneath this general result. Whereas females’ relative levels of satisfaction are in
line with the whole sample, for males the higher difference in job satisfaction relies on holding
a master´s versus a bachelor´s degree. On the other hand, women report a similar level of job
satisfaction when they have achieved short-cycle tertiary education or when they hold a
bachelor´s degree.
Men
Ve ry sati sfied
Satisfie d
Not very sat isfie d
Not at all satisfie d
BWV*
∆BWV*
Doctorate
0.4136 0.4955 0.0773 0.0136 1.3557
Mast e r
0.3816 0.5265 0.0776 0.0142 1.2798 0.0759
Bachelor or equivalent
0.3223 0.5614 0.0960 0.0203 1.1134 0.1664
Short-cycle tertiary education
0.2922 0.5702 0.1206 0.0170 1.0272 0.0862
Post-secondary non-tertiary
0.2412 0.6111 0.1188 0.0288 0.9244 0.1028
Upper secondary education
0.2140 0.6145 0.1399 0.0316 0.8522 0.0722
Lower secondary education
0.1972 0.5807 0.1716 0.0506 0.7664 0.0859
Primary education
0.1575 0.5888 0.1943 0.0595 0.6810 0.0853
Coe f. Va r 23.97
Women
Ve ry sati sfied
Satisfie d
Not very sat isfie d
Not at all satisfie d
BWV*
∆BWV*
Doctorate
0.3918 0.4948 0.0928 0.0206 1.3007
Mast e r
0.3177 0.5669 0.0946 0.0207 1.1422 0.1585
Bachelor or equivalent
0.2937 0.5798 0.1072 0.0193 1.0797 0.0625
Short-cycle tertiary education
0.2901 0.5805 0.1067 0.0227 1.0676 0.0121
Post-secondary non-tertiary
0.2608 0.5882 0.1289 0.0222 0.9855 0.0822
Upper secondary education
0.2316 0.6055 0.1357 0.0271 0.9210 0.0644
Lower secondary education
0.2237 0.5695 0.1582 0.0486 0.8484 0.0727
Primary education
0.1478 0.5623 0.2165 0.0733 0.6549 0.1935
Coe f. Va r 19.64
15
Figure 5: ∆BWV* components of job satisfaction by education level and gender
In summary, these results prove that:
1) job satisfaction increases with the level of education for the whole sample
2) this result holds when we consider each gender separately
3) the level of education can be considered as a plausible variable to explain job
satisfaction
In other words, it does not matter how we approach the evaluation of job satisfaction for
different education levels because we always obtain the same robust result: workers with higher
level of education report a higher degree of job satisfaction.
4.5 Job satisfaction and sectors
As reported by Heywood et al. (2002) and Ghinetti (2007) working in the public sector is related
to higher levels of job satisfaction. Ghinetti (2007) indicates that public employees differ from
private employees in the way they evaluate satisfaction with job security, consideration by
colleagues, and safety and health job features.
The EWCS provides information on the self-reported degree of satisfaction of workers in the
private sector, the public sector, in joint private-public organizations, and the not-for-profit
sector. In addition, a small percentage of workers reply that they work in a different sector which
the survey labels “other” but it does not provide more specific information about the economic
activities that these workers undertake in their jobs. Table 11 shows the number and proportion
of workers in each sector. As it can be expected, the large majority of workers, approximately
92%, work either in the private or the public sectors, being the proportion of workers in the
private sector almost three times higher than in the public sector.
Table 11: Number of workers and its percentage by sector
Sector
Total % Total
The private sector
29898 68.74
The public sector
10075 23.17
A joint private-public organisation
1404 3.23
The not-for-profit sector or an NGO
507 1.17
Oth er
1307 3.01
DK
301 0.69
16
Table 12 shows the BWV for workers in the different sectors.
Table 12: Job satisfaction by sector and BWV
Our first results show higher job satisfaction of workers in the not-for-profit or NGOs, in joint
private-public organizations and in the public sector than in the private sector. The category
denoted as ‘other’ ranks last. In addition, we have analyzed whether these results hold for both
genders. Table 13 and Table 14 show the BWV for males and females working in the different
economic sectors.
Table 13: Job satisfaction by sector and BWV for men
Table 14: Job satisfaction by sector and BWV for women
Let us discuss what we find with our evaluation method. First, job satisfaction is highest for
workers employed by not-for-profit institutions and nongovernment organizations (NGOs).
Secondly, in line with previous research, job satisfaction is higher in the public sector than in the
private one. These two findings hold for both men and women. Nevertheless, a closer analysis
of the differences among the components of the BWV allows us to make additional remarks
regarding the relationship between working in alternative sectors and the workers satisfaction.
The dispersion of the BWV components that are listed in Tables 14, 15 and 16 is the following
one: the coefficient of variation for women is 12.16% and 17.19% for men, while this measure
for all workers equals 14.53%. Thus, male workers show a higher level of dispersion in job
satisfaction across sectors than females.
Figure 6 illustrates the ∆BWV calculated in Tables 14, 15 and 16. The number in the horizontal
axis denotes the difference between the not-for-profit sector or NGO and a joint private-public
organization (1) a joint private-public organization and the public sector (2), the public sector
and the private sector (3), and the private sector and other (4).
Working for the public sector or for a joint private-public organization are related to similar
levels of job satisfaction for both men and women and thus for the whole sample. Women report
to be relatively more satisfied in NGOs and non-profit organizations than in the public sector.
The highest differences, however, for both males and females rest on working for the private
sector versus working in what the EWCS identifies as the “other” sector.
Sector
Ve ry sati sfied
Satisfied
Not v ery sat isfie d
Not at all satisfied
BWV*
∆BWV*
The not-for-profit sector or an NGO
0.3254 0.5385 0.1124 0. 0237 1.1560 -
A joint private-public organisation
0.2778 0.5840 0.1090 0. 0292 1.0621 0.0939
The public sector
0.2667 0.5968 0.1158 0. 0206 1.0482 0.0139
The private sector
0.2437 0.5899 0.1349 0. 0315 0.9634 0.0848
Othe r
0.2073 0.5210 0.2005 0. 0712 0.7704 0.1930
Men
Ve ry sati sfied
Satisfied No t ver y satisfi ed Not at all satisfied
BWV*
∆BWV*
The not-for-profit sector or an NGO
0.3136 0.5444 0.1243 0.0178 1.1426 -
A joint private-public organisation
0.2914 0.5714 0.1018 0.0354 1.1016 0.0409
The public sector
0.2727 0.5911 0.1133 0.0229 1.0741 0.0276
The private sector
0.2375 0.5949 0.1362 0.0315 0.9662 0.1078
Othe r
0.1944 0.4902 0.2320 0.0833 0.7155 0.2507
Women
Ve ry sati sfied
Satisfied No t ver y satisfi ed Not at all satisfied
BWV*
∆BWV*
The not-for-profit sector or an NGO
0.3314 0.5355 0.1065 0.0266 1.1580 -
The public sector
0.2628 0.6005 0.1175 0.0192 1.0285 0.1294
A joint private-public organisation
0.2640 0.5968 0.1162 0.0230 1.0259 0.0027
The private sector
0.2511 0.5839 0.1334 0.0316 0.9652 0.0606
Othe r
0.2187 0.5482 0.1727 0.0604 0.8224 0.1428
17
Figure 6: ∆BWV* components of job satisfaction by sector and gender
Finally, we have compared job satisfaction for men and women in each sector (see Table 15) and
found that women´s job satisfaction is higher than men´s when women work in the not-for-
profit sector and in the private sector whereas men are relatively more satisfied than women in
their jobs when they work in the public or in the joint private-public sectors. Nevertheless, the
coefficient of variation of the BWV between men and women working in the public and the
private sectors takes on low values, indicating that their satisfaction at work is similar. Notice
that there is a slight advantage of working in the private sector for females whereas males are
more satisfied than women in the public sector.
Table 15: BWV for job satisfaction by gender and sectors (separate evaluation)
In order to draw more light on this issue, we have calculated the BWV for the pooled sample
(see Table 16). The results are similar to the ones obtained previously, in both the ranking and
the intensities (the coefficient of variation of Table 16 is 14.05%). Working for the not-for-profit
sector ranks first, followed by the public and the private sectors.
Table 16: BWV for job satisfaction by gender and sectors (joint evaluation)
The not-for-profit sector
The public sector
Joint private-public
The private sector
Oth er
Men 0.9801 1.00838 1. 02325 0.98674 0.91675
Women 1.0199 0.99162 0.97675 1.01326 1.08325
Coef. Var 2.8143 1.1851 3.2880 1.8752 11.7733
BWV
∆BWV*
The not-for-profit sector or an NGO (W)
1.1728 -
The not-for-profit sector or an NGO (M)
1.1275 0.0453
A joint private-public organisation (M)
1.0872 0.0404
The public sector (M)
1.0595 0.0277
The public sector (W)
1.0427 0.0168
A joint private-public organisation (W)
1.0399 0.0028
The pri vate se ct or (W)
0.9784 0.0614
The pri vate se ct or (M )
0.9529 0.0256
Oth er (W )
0.8337 0.1192
Oth er (M )
0.7055 0.1282
18
In summary, we can observe the same pattern as before. Workers employed in the not-for-profit
and NGO organizations report the highest levels of job satisfaction, followed by workers
employed by the public sector (notice the higher satisfaction of men than women in the public
sector) and finally for workers employed by the private sector (where women seem to be more
satisfied than men).
We have also checked whether the relationship between job satisfaction and age was similar for
all sectors of the economy. Table 17 provides the BWV for the different age groups whose
frequencies appear in Appendix A1. As Table 17 and Figure 7 show, the behavior of the BWV for
different age groups varies widely across sectors. Besides, we cannot make a general statement
on the U-shaped relationship between job satisfaction and age since this seems to be true only
for workers who are employed at joint private-public organizations. It is indeed striking the
opposite feelings of workers of different ages employed in the not-for-profit and the private
sectors.
Table 17: BWV by sectors and age
Figure 7: BWV of different sectors for age intervals
4.6 Job satisfaction and working time
Another topic which has deserved our attention is the relationship between job satisfaction and
working hours. Are more satisfied workers who work full-time or part-time? Table 18 shows the
BWV calculations.
Group Total NGO Joint Public Private
Coef. Var.
< 25 1.0476 0.9587 1.1140 1.0148 1.1593 79.42
26 - 35 0.9787 0.9974 0.9490 1.0360 0.9748 32.28
36-45 0.9420 1.1302 0.9886 1.0056 0.9496 75.76
46-55 0.9380 0.9396 0.9462 0.9341 1.0509 50.03
56-65 0.9530 0.9793 0.9373 1.0302 0.9425 38.12
> 65 1.1408 0.9949 1.0649 0.9793 0.9228 84.16
Coef. Var.
79.94 67.47 73.18 38.02 89.90
19
Table 18: Job satisfaction by workday duration and BWV
Workday duration does not appear to have much influence in job satisfaction since the
component of the BWV for people working full-time and part-time is very similar: the difference
is approximately 2% higher for workers employed full-time than for part-time work.
As it is well known, men and women follow different strategies in their work-life balance, with
women assigning more time to household work every day and thus choosing part-time jobs.
Indeed, the proportion of men and women working full and part-time in the 2015 EWCS wave is
the following:
Table 19: Distribution of workday duration between genders
Thus, it is mostly women who choose or decide, for whatever reason, to work part-time. Let us
now calculate the BWV for men and women working full-time and part-time. Tables 22 and 23
show the results.
Table 20: Job satisfaction for full time by gender and BWV
As it happens with the whole sample, when we consider full-time work, gender does not induce
large differences in work satisfaction (the component of the BWV for women is only 1% higher
than the one for men).
Table 21: Job satisfaction for part time by gender and BWV
On the other hand, when we consider part-time work, we find larger differences between men
and women. In this case, the component of the BWV for women is about 14% higher than the
one for men. In part-time jobs, women report a higher level of job satisfaction than men.
Similar results have been found in previously published articles. It has been argued that women
value flexibility over other aspects of jobs in order to be able to accommodate to family life and
household work (Bender et al., 2005). Booth and van Ours (2013) find that women in part-time
work have high levels of job satisfaction, a low desire to change their working hours, and mainly
live in families where household production is highly gendered. Their results suggest that part-
time jobs are what most Dutch women want. Furthermore, without the possibility of taking part-
time jobs, female labor force participation could be substantially lower because many women
facing the choice between a full-time job and not working at all could choose the latter.
Let us continue with a joint evaluation of men and women working full-time and part-time.
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
Full time 0.2460 0.6017 0.1263 0.0260 1.0098
Part time 0.2660 0.5478 0.1462 0.0399 0.9902
Full time Part time
Men 0.5566 0.3085
Women 0.4433 0.6913
Full time
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
Men 0.2448 0.6012 0.1268 0.0272 0.9947
Women 0.2474 0.6024 0.1257 0.0245 1.0053
Part time
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
TOTAL
Men 0.2446 0.5317 0.1726 0.0511 0.9343
Women 0.2755 0.5550 0.1345 0.0349 1.0657
20
Table 22: BWV for job satisfaction when work time duration and gender are jointly evaluated
Table 22 shows that when working time and gender are jointly evaluated, women report higher
levels of job satisfaction than men and it is working part-time which makes them be more
satisfied in the labor market whereas men report higher levels of satisfaction working full-time
than part-time.
4.7 Job satisfaction and employment status
In the EWCS 82.45% of surveyed individuals are employees under different types of contracts
whereas 17.55% of workers identify themselves as self-employed. We now focus on the
relationship between job satisfaction and employment status and thus proceed to analyze
whether being an employee or self-employed leads to different levels of job satisfaction. Table
23 shows that employees are more satisfied than self-employed workers.
Table 23: Job satisfaction for employment status and BWV
If we take the worker´s gender into account (see Table 24), the ranking is the following one:
women who are employees rank first, followed by men who are employees and in third and
fourth places we find self-employed women and self-employed men.
Table 24: Job satisfaction for employment status and gender and BWV joint evaluation
Table 25: BWV for job satisfaction, employment regime and gender, independent evaluation
In Table 24 we show the calculations of BWVs when we undertake independent evaluations. The
new calculations corroborate the previous results: being an employee reports higher job
satisfaction for both men and women. Nevertheless, as it can be seen in Tables 25, 26 and 27,
the worker’s employment status does not seem to cause large differences in job satisfaction.
BWV
Women (P) 1.0434
Women (F) 1.0274
Men (F) 1.0166
Men (P) 0.9126
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
Empl oyee
0.2521 0.5865 0.1314 0.0300 1.0143
Self-employed
0.2403 0.5927 0.1348 0.0322 0.9857
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
BWV
Employee (M)
0.2477 0.5873 0.1342 0.0308 1.0013
Self-employed (M)
0.2342 0.5963 0.1339 0.0356 0.9713
Employee (W)
0.2566 0.5856 0.1286 0.0291 1.0277
Self-employed (W)
0.2461 0.5892 0.1357 0.0290 0.9997
Men Women
Empl oyee Self-employed
Empl oyee
1.0152 1.0139 Men 0.9870 0.9856
Self-employed
0.9848 0.9862 Wome n 1.0130 1.0144
Coef. Variation
2.15 1.96 1.84 2.03
21
4.8 Job satisfaction and type of contract
In this section we focus on the relationship between the labor contract that each employee has
and her degree of self-reported job satisfaction. Table 26 shows the number and percentage of
individuals who hold different types of contracts in the EWCS:
Table 26: Prevalence of different types of labor contract in the EWCS
The most common type of contract is the indefinite one followed by fixed-term contracts and
‘other’ contracts. This last category encloses heterogeneous types of labor contracts across the
countries surveyed in the EWCS.
Table 27: Distributions, BWV* and ∆BWV* of job satisfaction by type of labor contract
Table 27 shows the distribution of self-reported job satisfaction for different labor contracts
ranked by their BWV value. Tables 30 and 31 show the same calculations for each gender. Tables
29, 30 and 31 have in common the relatively large first differences of the BWV components of
the apprenticeship/training contracts. These results suggest that individuals holding an
apprenticeship/training contract report higher levels of job satisfaction independently of their
gender.
Table 28: Distributions, BWV* and ∆BWV* of job satisfaction by type of contract for men
Table 29: Distributions, BWV* and ∆BWV* of job satisfaction by type of contract for women
This conclusion is reinforced by the fact that the coefficients of variation decrease from 23.91%
for the whole sample, 24.20% for men and 23.95% for women to 10.40%, 9.96% and 9.03%
respectively when we do not consider the apprenticeship/training contracts. Moreover, we
observe two additional regularities in Tables 29, 30 and 31: the relative BWV-dominance of the
Type of contract Total % Total
Indefinite
27281 76.18
Fixed term c ontract
4169 11.64
Other
3612 10.09
Temporary employment agency
506 1.41
Apprenticeship/training
242 0.68
Whole sample
Very satisfied
Satisfied
Not very satisfied
Not at all satisfied
BWV*
∆BWV*
Apprenticeship/training 0.3760 0. 5496 0.0620 0. 0124 1.4034 -
Indefinite 0.2502 0. 6092 0.1172 0. 0233 1.0272 0. 3762
Fixed term cont ract 0.2243 0.5781 0.1593 0. 0384 0.9018 0.1254
Other 0. 2234 0.5449 0.1750 0.0568 0.8522 0.0496
Temporary employment agency 0.1996 0.5632 0.1917 0. 0455 0.8154 0.0368
Men
Ve ry sati sfied
Satisfied
Not very sat isfie d
Not at all satisfied
BWV*
∆BWV*
Apprenticeship/training
0.3876 0.5116 0.0853 0. 0155 1.4045 -
Indefinite
0.2421 0.6160 0.1174 0. 0244 1.0355 0.3690
Fixed term cont ract
0.2223 0.5718 0.1662 0. 0396 0.9100 0.1255
Temporary employment agency
0.2099 0.5420 0.1870 0. 0611 0.8327 0.0773
Other
0.2008 0.5485 0.1909 0. 0598 0.8174 0.0153
Women
Ve ry sati sfied
Satisfied
Not very sat isfie d
Not at all satisfied
BWV*
∆BWV*
Apprenticeship/training
0.3628 0.5929 0.0354 0. 0088 1.4050 -
Indefinite
0.2576 0.6030 0.1170 0. 0224 1.0169 0.3881
Fixed term cont ract
0.2262 0.5832 0.1533 0. 0373 0.8914 0.1255
Other
0.2465 0.5411 0.1587 0. 0537 0.8899 0.0015
Temporary employment agency
0.1893 0.5844 0.1975 0. 0288 0.7967 0.0932
22
indefinite contract with respect to the others and the relative similarity of the BWV values for
fixed-term, temporary-employment agency and other contracts.
We have gone a step further to investigate whether gender plays an important role in this issue
and have found further empirical evidence to support the gender/job satisfaction paradox:
under any type of labor contract, females are more satisfied than their male counterparts. Table
30 shows this result. Nevertheless, although Table 30 shows that women fare better than men
under all types of labor contracts, their differences do not appear to be relevant. In other words,
gender does not seem to be important to explain differences in job satisfaction due to the type
of contract.
Table 30: BWV* and ∆BWV* of job satisfaction for gender and type of contract
5 Conclusions
Along with the relatively recent line of research on happiness, job satisfaction has become
increasingly important in the Economics Literature because of its expected -and proved- effects
on most aspects of workers lives. Authors have mainly focused on both the determinants and
implications of job satisfaction following an empirical approach. Data on job satisfaction have
been drawn from surveys made to workers in which they are asked to report their degree of
satisfaction on a categorical scale. Results have often been striking and controversial. In this
paper, we have reevaluated previous findings using a new methodology which has been
specifically designed to analyze distributions of categorical data. Our main conclusions are the
following ones:
1. Most workers report to be satisfied or very satisfied with their jobs.
2. Countries can be ordered in terms of the job satisfaction of their workers. Among 35
countries, Denmark is ranked first while Albania appears in the last place. In general,
workers in Scandinavian countries and Northern Europe report the highest levels of job
satisfaction, followed by countries in Continental Europe, the Southern Mediterranean
countries and finally, Eastern European countries. Both the variability of the BWV for
the different countries and the emergence of several clusters where job satisfaction is
perceived in a similar fashion lead us to conclude that the “country factor” is relevant
explaining job satisfaction.
3. Women show higher degrees of job satisfaction than men, supporting the gender/job
satisfaction paradox.
4. Job satisfaction is U-shaped in age. However, there are slight differences between
genders and when we consider the sector in which the workers are employed, we find
very different patterns in this relationship.
5. For both female and male workers, either jointly or separately evaluated, job
satisfaction is increasing on the level of education achieved by the worker.
6. Workers who are employed in the not-for-profit and NGO sectors are more satisfied in
their jobs than other workers. They are followed by workers employed in the public
sector and in third place for workers employed in the private sector.
7. In terms of job satisfaction, working full-time is only slightly better than working part-
time. Women are more satisfied than men with any working hours arrangement but the
Fixed term Temporary Training Indefinite Othe r
Women 1.0148 1.0084 1.0121 1.0154 1.0641
Men 0.9853 0.9916 0.9879 0. 9846 0.9359
∆BWV* 0.0295 0.0167 0.0243 0.0308 0.1282
23
difference in job satisfaction between males and females is much higher when they both
work part-time: whereas women seem to be satisfied working part-time, their male
counterparts report to be substantially less satisfied in part-time jobs.
8. Employees report to be more satisfied in their jobs than workers who are self-employed.
Again, women, either employed or self-employed, feel more satisfied than men.
9. Having an indefinite labor contract provides a higher level of satisfaction than having a
fixed-term contract. However, those who report to be most satisfied are apprentices
and workers on training whereas the least satisfied are workers on temporary
employment hired by an employment agency.
All the above are univocal and unambiguous findings that arise from applying the Balance Worth
procedure to the last wave of the European Working Conditions Survey. We hope that our
results can be used to shed new light upon the most controversial issues that have been
addressed by previous empirical research on job satisfaction.
References
Ahn, N. and J.R. García (2004). Job Satisfaction in Europe. Documento de Trabajo 2004-16.
FEDEA.
Bender, K. A., Donohue, S. M. and J.S. Heywood (2005). Job satisfaction and gender segregation.
Oxford Economics Papers 57: 479-496.
Blanchflower, D.G. and A.J. Oswald (2004). Well-being over time in Britain and the USA. Journal
of Public Economics, 88: 1359-1386.
Blanchflower, D.G. and A.J. Oswald (2011). International Happiness: A New View on the Measure
of Performance. Academy of Management Perspectives, 25 (1): 6-22.
Booth, A.L and J.C. van Ours (2013). Part-time jobs: what women want?, Journal of Population
Economics, 26:263-283.
Carsten, J.M. and P.E. Spector (1987). Unemployment, job satisfaction, and employee turnover:
A meta-analytic test of the Muchinsky model. Journal of Applied Psychology, 72, 374-379.
Clark, A.E. (1997). Job satisfaction and gender: Why are women so happy at work?, Labour
Economics, 4, 341-372.
Clark, A.E. and A.J. Oswald (1996). Satisfaction and Comparison Income. Journal of Public
Economics, 61, 359-381.
Clark, A.E, Oswald, A.J. and P. Warr (1996). Is Job Satisfaction U-Shaped in Age?, Journal of
Occupational and Organizational Psychology, 69, 57-81.
European Foundation for the Improvement of Living and Working Conditions. (2018). European
Working Conditions Survey Integrated Data File, 1991-2015. [data collection]. 7th Edition. UK
Data Service. SN: 7363, http://doi.org/10.5255/UKDA-SN-7363-7
Fernández Macías, E. and R. Muñoz de Bustillo (2014). Job satisfaction. Encyclopedia of Quality
of Life and Well-Being Research. Editors: Michalos, Alex C. (Ed.), 3451-3454.
24
Fiorillo, D. and N. Nappo (2014). Job satisfaction in Italy: Individual characteristics and social
relations. International Journal of Social Economics, 41 (8), 683-704.
Freeman, R.B. (1978). Job Satisfaction as an Economic Variable. American Economic Review,
68(2), 135-141.
García-Mainar, I. and V. M. Montuenga-Gómez (2019). Over-Qualification and the Dimensions
of Job Satisfaction. Social Indicators Research. https://doi.org/10.1007/s11205-019-02167-z
Gaziougly, S. and A. Tansel (2006). Job satisfaction in Britain: Individual and job related factors.
Applied Economics, 38, 1163–1171.
George, J.M. and G.R. Jones (2008). Understanding and managing Organizational Behavior (Fifth
Edition). Upper Saddle River: New Jersey, Pearson Prentice Hall.
Ghinetti, P. (2007). The Public-Private Job Satisfaction Differential in Italy. Labour, Vol. 21, No.
2, pp. 361-388.
Hamermesh, D. (1977). Economics for Job Satisfaction and Worker Alienation, in Orley
Ashenfelter and Wallace Oates, eds., Essays in Labor Market and Population Analysis, Princeton.
Herrero, C. and A. Villar (2013). On the comparison of group performance with categorical data.
PLoS ONE, 8 (12), e84784. https://doi.org/10.1371/journal.pone.0084784.
Herrero, C. and A. Villar (2018). The balanced worth: A procedure to evaluate performance in
terms of ordered attributes. Social Indicators Research, 140: 1279-1300.
Heywood, J.S., Siebert, W.S. and X. Wei (2002). Worker Sorting and Job Satisfaction: The Case of
Union and Government Jobs. ILR Review, 55(4), 595-609.
Hinz, A., Häuser, W., Glaesmer, H. and E. Brähler (2016). The relationship between perceived
own health state and health assessments of anchoring vignettes. International Journal of Clinical
and Health Psychology 16, 128-136.
Johnson, G. and W. Johnson (2000). Perceived Overqualification and Dimensions of Job
Satisfaction: A Longitudinal Analysis. The Journal of Psychology Interdisciplinary and Applied
134(5), 537-55.
Lieberson, S. (1976). Rank-sum comparisons between groups. Sociological Methodology, 7: 276-
291.
Muñoz de Bustillo, R. and E. Fernández-Macías (2005). Job satisfaction as an indicator of the
quality of work. Journal of Socio-Economics, 34 (5), 656-673.
Oshagbemi, T. (2000). Gender differences in the job satisfaction of university teachers, Women
in Management Review, 15(7), 331-343.
Osterman, P. (2013). Introduction to the Special Issue on Job Quality: What Does It Mean and
How Might We Think About It? ILR Review, 66(4), Special Issue on Job Quality, 739-752.
Oswald, A.J. (1997). Happiness and Economic Performance. Economic Journal, 107, 1815-1831.
Pichler, F. and C. Wallace (2009). What Are the Reasons for Differences in Job Satisfaction across
Europe? Individual, Compositional, and Institutional Explanations. European Sociological
Review, 25(5), 535-549.
25
Rode, J.C. (2004). Job satisfaction and life satisfaction revisited: A longitudinal test of an
integrated model. Human Relations, Volume 57(9): 1205–1230.
Sloane, P.J. and H. Williams (2000). Job Satisfaction, Comparison Earnings, and Gender, Labour,
vol. 14(3): 473-502.
Sousa-Poza, A. and A.A. Souza-Poza (2000). Well-being at work: a cross-national analysis of the
levels and determinants of job satisfaction, Journal of Socio-Economics, 29, 517-538.
Sousa-Poza, A. and A.A. Souza-Poza (2003). Gender differences in job satisfaction in Great
Britain, 1991-2000: permanent or transitory? Applied Economics Letters, 10:11, 691-694.
Vila, L.E. and B. García-Mora (2005). Education and the determinants of job satisfaction.
Education Economics, 4, 409-425.
Warr, P. (1999). Well-Being and the Workplace. In D. Kahneman, E. Diener, & N. Schwarz (Eds.),
Well-Being: The Foundations of Hedonic Psychology (pp. 392-412). New York: Russell Sage
Foundation.
World Happiness Report (2015). The Earth Institute. Columbia University.
26
Appendix
A1: Frequencies for the different age groups in each sector of Table 8.
NGO
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
< 25 0.2667 0.6444 0.0444 0.0444
26 - 35 0. 3300 0.5200 0.1400 0.0100
36-45 0.3922 0.4706 0.1275 0.0098
46-55 0.3167 0.5083 0.1500 0.0250
56-65 0.2747 0.6374 0.0659 0.0220
> 65 0.3333 0.5208 0.0833 0.0625
Joint
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
< 25 0.3421 0.5439 0.0789 0.0351
26 - 35 0. 2734 0.5843 0.0974 0.0449
36-45 0.2890 0.5751 0.1105 0.0255
46-55 0.2637 0.5989 0.1181 0.0192
56-65 0.2607 0.6000 0.1143 0.0250
> 65 0.3333 0.5238 0.1429 0.0000
Public
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
< 25 0.2688 0.6090 0.0978 0.0244
26 - 35 0. 2797 0.5970 0.1114 0.0120
36-45 0.2852 0.5673 0.1184 0.0292
46-55 0.2333 0.6288 0.1181 0.0198
56-65 0.2872 0.5777 0.1163 0.0187
> 65 0.2558 0.6105 0.1221 0.0116
Private
Ve ry sati sfie d
Satisfied
Not very sat isfie d
Not at all satisfied
< 25 0.2918 0.5881 0.1004 0.0197
26 - 35 0. 2303 0.6039 0.1325 0.0333
36-45 0.2273 0.5921 0.1479 0.0328
46-55 0.2654 0.5796 0.1243 0.0307
56-65 0.2257 0.5911 0.1473 0.0359
> 65 0.2378 0.5498 0.1792 0.0332
