Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

~- "~ ." : . .

LABOUR
ECONOMICS
ELSEVIER Labour Economics 4 (1997) 341-372

Job satisfaction and gender: Why are women so


happy at work?
A n d r e w E. Clark *
OECD, Education, Employment, Labour & Social Affairs, 2 rue Andrd Pascal, 75775 Paris Cedex 16.
France
Received 24 April 1995; accepted 17 October 1996

Abstract

By most objective standards, women's jobs are worse than men's, yet women report
higher levels of job satisfaction than do men. This paper uses a recent large-scale British
survey to document the extent of this gender differential for eight measures of job
satisfaction and to evaluate the proposition that identical men and women in identical jobs
should be equally satisfied. Neither the different jobs that men and women do, their
different work values, nor sample selection account for the gender satisfaction differential.
The paper's proposed explanation appeals to the notion of relative well-being, especially
relative to workers' expectations. An identical man and woman with the same jobs and
expectations would indeed report identical job satisfaction, but women's expectations are
argued to be lower than men's. This hypothesis is supported by the finding that the gender
satisfaction differential disappears for the young, the higher-educated, professionals and
those in male-dominated workptaces, for all of whom there is less likely to be a gender
difference in job expectations. © 1997 Elsevier Science B.V.

JEL classification: C13; J16; J28; )71

Keywords: Job satisfaction; Gender; Comparisons; Expectations

1. Introduction
One o f the most consistent findings in empirical labour economics is that o f a
large and significant difference b e t w e e n m e n ' s and w o m e n ' s pay. Estimates of the

* Tel.: +33-1-45249279; fax: +33-1-45241968; e-maih andrew.clark@oecd.org.

0927-5371/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved.


PH S 0 9 2 7 - 5 3 7 1 (97)0001 0-9
342 A.E. Clark/Labour Economics 4 (1997) 341-372

ratio of the pay of women to that of equivalently qualified men doing the same
type of job have ranged from 94 to 49%, with a recent estimate of 80% 1. There is
also extensive evidence that women's jobs are worse than men's in terms of hiring
and firing, job content, promotion opportunities and sexual harassment e. Even so
and despite their higher levels of reported stress in their life (see Argyle, 1989 and
Clark and Oswald, 1994), women consistently report higher job satisfaction scores
than do men 3. This paper considers the implications of this gender job satisfaction
differential. It argues that the difference in responses represents a real difference in
utility from working 4, but that, objectively, women's jobs are worse than men's.
The resolution of this paradox is suggested to lie in the importance of expectations
in well-being: those who expect less from working will be more satisfied with any
given job. This finding warns against the direct translation from objective charac-
teristics to subjective well-being. In this case, rather than indicating that women's
jobs are more attractive than men's, women's higher job satisfaction reflects their
lower expectations, which themselves likely result from the poorer position in the
labour market that women have held in the past. As the difference between men's
and women's jobs is eroded, so will be the gender difference in expectations and
job satisfaction (once relevant job characteristics have been controlled for).
Section 2 discusses the use of job satisfaction data as a measure of individual
well-being in economics, and presents four explanations of why the responses to
survey measures of well-being may differ by gender: because individuals and their
jobs are different; because men and women want different things from a job;
because sample selection is at work and because job outcomes are evaluated
relative to expectations. Section 3 then uses data from the British Household Panel
Survey (BHPS) to document the extent of this gender differential for eight
different types of job satisfaction, and evaluates the first of these explanations via
the introduction of a wide array of individual and job characteristics as control
variables in ordered probit regressions.
Sections 4 - 6 in turn test the remainder of Section 2's explanations for three
measures of job satisfaction: satisfaction with pay, satisfaction with the work itself
and overall job satisfaction. Section 4 analyses the information on work values
included in the BHPS. These work values are shown to differ markedly between

1 See Wright and Ermisch (1991) for the source of these estimates; also Blau and Kahn (1992) and
Equal Opportunities Commission (1992).
2 See, for example, Brown Johnson et al. (1992), Hakim (1991), Lewis (1992), Loprest (1992),
Riach and Rich (1987) and Townshend-Smith (1989).
3 Women's higher job satisfaction has been reported in recent work by Blanchflower and Oswald
(1992), Blanchflower et al. (1993), Meng (1990) and Woittiez and Theeuwes (1994), using British,
American, Canadian and Dutch data, respectively. Curtice (1993) finds the same relationship in 1989
ISSP data covering eleven countries.
4 This paper does not consider biological or genetic explanations for gender differences in behaviour
or evaluations; these are discussed by Arvey et al. (1989), Birke (1986), Garai and Scheinfeld (1968),
Moir and Jessel (1989), Rose et al. (1984) and Watson and Lindgren (1973).
A,E. Clark~Labour Economics 4 (1997) 341-372 343

the sexes, but, while they are very strongly correlated with job satisfaction, they
do not account for the gender satisfaction differential.
It is well-known that women's participation rates are lower than men's. If
participation is correlated with the prospective satisfaction of working, then
women's higher satisfaction could have a simple statistical, rather than theoretical,
explanation. Section 5 considers this proposition by the use of Heckman sample
selection correction in OLS regressions. The results do not support the self-selec-
tion explanation of the gender job satisfaction differential.
Some recent research into well-being has argued that certain of its arguments
may be evaluated in a relative, rather than absolute, fashion and that expectations
may form an important part of the reference level. Section 6 investigates this
hypothesis in two ways. First a measure of comparison income, which may pick
up income expectations, is introduced into the empirical analysis: this specification
results in a somewhat reduced, but still significant, estimated female-male satis-
faction differential. Second, various interaction experiments show that the gender
satisfaction differential disappears for young workers, highly-educated workers,
workers in professional occupations and workers in male-dominated workplaces,
for which groups there is less likely to be a gender difference in job expectations.
Last, Section 7 considers the separate determinants of job satisfaction for men
and women and Section 8 presents some concluding comments.

2. Theory

Recent years have seen an increase in economists' interest in the empirical


analysis of subjective well-being variables. One such variable is job satisfaction, in
its various forms. Following Clark and Oswald (1996), this can be though of as a
type of sub-utility function, u, representing utility from working in an overall
utility function, v ~ v(u, ~), where/~ is utility from other areas of life. The utility
from working is usually considered to be of the form
u = u( y, h, i, j ) , (1)
where y is income, h is hours of work, and i and j are sets of individual and
job-specific characteristics, respectively. The empirical analysis of job satisfaction
responses by groups has turned up a number of common correlations: for example,
women, the lower-educated and the young or old report significantly higher levels
of job satisfaction than do men, the higher-educated and the middle-aged, respec-
tively.
The analysis of these correlations is useful for several reasons. First, job
satisfaction is, according to Argyle (1989), one of the three most important
predictors of overall well-being (the other two are marriage and family satisfac-
tion) and the distribution of well-being is one of the primary concerns of social
science. Second, job satisfaction turns out to be correlated, in the expected
direction, with worker behaviour. Akerlof et al. (1988), McEvoy and Cascio
344 A.E. Clark/Labour Economics 4 (1997) 341-372

(1985) and Freeman (1978) all find that job satisfaction predicts future quits 5,
while Clegg (1983) and Mangione and Quinn (1975) show that job satisfaction
responses are correlated with absenteeism (negatively) and worker productivity
(positively), respectively. The understanding of workers' subjective well-being
thus provides an additional route towards the understanding of certain important
labour market behaviours. Third, job satisfaction may be as close as we are likely
to come to a proxy measure of utility at work, upon which a great deal of
microeconomics is based. Job satisfaction data may then enable us to specify more
closely the likely structure of the ubiquitous utility function.
All of the above relies on satisfaction data being comparable across individuals:
does one person use the same scale of answers to job satisfaction questions (which
are usually on a scale of one to four, five or seven, with higher numbers indicating
greater contentment) in the same way as another? The answer is undoubtedly no,
but there does seem to be signal mixed in with the noise. If answers were purely
idiosyncratic, then none of the correlations between labour market behaviour and
satisfaction scores mentioned above would have been found. In addition, it is not
clear why whole groups, such as women or older workers, should systematically
understand the satisfaction scale so differently 6. Last, psychologists and sociolo-
gists have been using such data for many years and have repeatedly validated
them. Their view, which is also that of this paper, is that there is useful
information contained in cross-section answers to questions on well-being.
This paper thus considers that female-male differences in responses to job
satisfaction questions reflect a real difference in well-being; it also considers that,
once all relevant variables are controlled for, there is no good reason why gender
should enter into the vector of i variables in Eq. (1), i.e. an identical man and
woman in identical jobs should report the same job satisfaction score. Much of this
paper will be concerned with what 'identical' means, testing four explanations of
the gender satisfaction differential and concluding that it includes more than just
objective job and individual characteristics.

2.1. Jobs and gender

The first explanation relies on men's and women's different personal and job
characteristics. Empirical research into job satisfaction has highlighted correlations

5 Freeman finds that, in two out of the three datasets analysed, job satisfaction is at least as good a
predictor of quits as are wages.
6 The assumption that women are less assertive than men in their use of language is critically
challenged by Crawford (1995). The issue of utility comparisons is discussed in Elster and Roemer
(1991), Harsanyi (1986) and Van Praag (1991). The data used in this paper do contain a measure of
mental stress, the GHQ-l 2 (Goldberg, 1972). As a test, this was added as an explanatory variable to the
regressions. The results showed that men and women with identical levels of mental stress, as well as
identical individual and job characteristics, stil! reported significantly different levels of job satisfac-
tion.
A.E. Clark~Labour Economics 4 (1997) 341-372 345

with a number of these, such as age, education, health, union membership,


establishment size and hours of work. As the distribution of many of these
characteristics is correlated with gender, the job satisfaction regression analysis in
Section 3 will control for a wide variety of both individual and job-related
characteristics, in addition to gender. As in the analysis of wage differentials, it is
important to compare individuals who are as close to alike as possible, apart from
their gender.

2.2. Work values

The individual variables which are associated with reported well-being need not
only be observable (age, education, income, etc.), they may also reflect an
individual's priorities or values. Section 4 thus controls for the values which the
individuals themselves report for different aspects of working; added to the
previous variables measuring individual and job characteristics, this specification
amounts to controlling for the i and j variables in Eq. (1).

2.3. Self-selection

Any non-random sample may produce correlations that result from sample
selection rather than from a structural relationship. For the problem at hand, as is
often the case in labour economics, the sample consists only of employees, and the
probability of being an employee is likely correlated with how the individual
would feel about working, i.e with their potential job satisfaction. The probable
relationship between the two is positive: the more satisfying the job would be, the
more likely the individual is to work. As fewer women than men work, it may be
that dissatisfied men are working whereas potentially dissatisfied women occupy
some other labour force status. This problem is dealt with in the same way as that
of estimating the gender wage gap: by controlling for the process of selection into
the sample of employees.

2.4. Relative utility and expectations

The final explanation rests upon a different specification of Eq. (1). It has long
been suggested in the social sciences that satisfaction or well-being may be partly
determined by relative rather than absolute arguments. This idea of comparisons
has its basis in the work of Duesenberry (1949) and has been developed outside of
economics by Adams (1963), Homans (1961), Stouffer et al. (1949) and Runciman
(1966). Income, for example, may be evaluated relative to some comparison level,
y*, so that
u=u(y, y*,h,i,j). (2)
The higher is y *, the lower is the individual's relative income and the lower is
their well-being or satisfaction; y * is thus negatively correlated with u.
346 A.E. Clark~Labour Economics 4 (1997) 341-372

Although this is a simple idea, it is difficult to test empirically because of


uncertainty about what y * represents. The economists at Leiden University, who
investigated such comparison effects in the Welfare Function of Income (the
amount of money that individuals assign to verbal labels such as 'excellent',
'good', 'adequate' and so on), typically assumed that y* was the average income
of people with the same age, education and sex as the respondent (Hagenaars,
1986; Kapteyn and Van Herwaarden, 1980). Clark and Oswald (1996) expanded
this idea by using a standard wage equation, based on a range of individual and
job characteristics, to predict a 'going rate' for each individual in their current job
(see also Hamermesh, 1977; Lrvy-Garboua and Montmarquette, 1994; Sloane and
Williams, 1995). Clark (1996b) has argued that the income of other household
members may help to determine y *, while Clark (1996a) examines the possibility
that past wages may inform the individual's evaluation of their current income and
thus their job satisfaction.
Comparisons may apply to aspects of the job other than income, for example
hours of work, stress, autonomy and authority. A general relative utility from work
function may therefore include a vector of comparison levels E,
u = u ( y , h, i, j , E ) . (3)
The comparison levels in E may come from observation of others, from one's
own experiences in the past, or from one's feelings of what one should receive i.e.
from expectations. This paper will concentrate on the part of E which concerns
expectations. The paradox of women' s higher satisfaction with worse jobs can thus
be resolved by women's lower expectations. Some evidence consistent with this
view will be discussed in Section 6.
In theoretical work, utility functions in which some arguments are evaluated
relatively have been proposed to resolve some of the inconsistencies of observed
behaviour with Von Neumann-Morgenstern utility functions. These reference
levels have variously been formulated as the individual's current level of the
argument, especially with respect to income or assets (prospect theory: Kahneman
and Tversky, 1979), the income or goods that could have been received had
uncertainty been resolved differently (regret theory: Bell, 1982; Loomes and
Sugden, 1982), or prior expectations (disappointment theory: Bell, 1985). It is this
latter formulation which is closest to our interpretation of Eq. (2) 7.

7 Kahneman and Tversky (1979) remark that "This assumption is compatible with basic principles
of perception and judgement... When we respond to attributes such as brightness, loudness, or
temperature, the past and present context of experience defines an adaption level, or reference point,
and stimuli are perceived in relation to this reference point" (p. 277). Bell (1985) motivates
disappointment theory with the following example: "Your boss tells you that he is delighted with your
performance over the past year and is giving you a $5,000 bonus. Are you pleased? If you were not
expecting a bonus, you will be delighted. If you were expecting a $10,000 bonus, you will be
disappointed. The satisfaction you feel with the bonus you are given will depend upon your prior
expectations" (p. 1).
A.E, Clark/Labour Economics 4 (1997) 341-372 347

3. Men, w o m e n and their jobs

The data used in this paper come from wave 1 of the British Household Panel
Survey (BHPS). This nationally representative survey interviewed nearly 10,000
adults in 5500 households over the period September to December 1991 and
collected a wide range of information concerning the household and its adult
members' health, employment, finances and opinions. Further details of this
survey are available in Buck et al. (1994).
The questions used to measure job satisfaction are as follows. Initially, individ-
uals were asked to rate their satisfaction levels with seven specific facets of their
job: promotion prospects, total pay, relations with supervisors, job security, ability
to work on their own initiative, the actual work itself and hours of work. Each of
these criteria was to be given a number from one to seven, where a value of one
corresponded to 'not satisfied at all' and a value of seven corresponded to
'completely satisfied'. Finally, individuals were asked "All things considered,
how satisfied or dissatisfied are you with your present job overall using the same
1 - 7 scale?"

Table 1
Type of job satisfaction Women Men All t-statistic on
gender difference
Mean reported job satisfaction levels
Overall 5.71 5.31 5.50 9.56
Promotion prospects 4.36 4.22 4.28 2.38
Total pay 4.65 4.33 4.48 5,87
Relations at work 5.86 5.47 5.66 8.36
Job security 5.41 5.01 5.20 7,39
Initiative 5.87 5.80 5.84 1.59
Actual work itself 5.78 5.56 5.67 5.08
Hours 5.56 5.08 5.31 9.79

Percentage highly satisfied


Overall 65.0 52.9 58.7 8.88
Promotion prospects 32.7 29.6 31.0 2.23
Total pay 38.7 31.2 34.8 5.70
Relations at work 69.5 60.8 65.0 6.48
Job security 59.1 50.2 54.5 6.40
initiative 70.5 68,6 69.5 1.53
Actual work itself 66.7 60,6 63.6 4.55
Hours 60.4 47.9 53.9 9.06

Note: The number of observations varies slightly with the satisfaction measure used. There are 2501
and 2693 observations on female and male workers respectively for overall job satisfaction. All
numbers refer to weighted data.
348 A.E. C l a r k ~ L a b o u r Economics 4 (1997) 341-372

Table 1 shows the reported levels of job satisfaction in the BHPS survey 8 The
first two columns in the top half of the table show the mean level of the
satisfaction score for women and men respectively. In addition, as job satisfaction
is ordinal rather than cardinal (i.e. job satisfaction of 6 is not twice as high as job
satisfaction of 3), the first two columns of the bottom half of the table show the
percentage of women and men respectively who say that they are 'highly satisfied'
(reporting satisfaction of 6 or 7 on the 1-7 scale). Column 4 reports the t-statistic
from the test of identical means by gender. Women report higher scores than do
men for all of the eight job satisfaction measures and for six of the measures the
differential is easily significant at the 0.1% level. The differential is smaller for
satisfaction with promotion prospects and satisfaction with initiative.
The first explanation of women's higher job satisfaction advanced in Section 2
was that many individual and job-related characteristics differ markedly by gender.
To test this, a wide variety of explanatory variables were included in a multivariate
analysis of all eight job satisfaction measures; Tables 2 and 3 report the results.
The ordinal dependent variable requires the use of ordered probit regressions
(Zavoina and McKelvey, 1975)9. The estimated coefficients on the 'male'
variable reveal that only for satisfaction with job security and satisfaction with
initiative do these controls account for the gender differential; for the other six
measures women still report significantly higher job satisfaction than do men.
Many of the control variables used in Tables 2 and 3 turn out to be important
predictors of satisfaction themselves m. There is a well-defined U-shaped relation-

8 All statistical work refers to employees only. The self-employed are asked similar questions, but
their low response rate with respect to income questions and the inapplicability of questions referring to
establishment size, union membership, temporary work and so on led to their exclusion from the
analysis. The job satisfaction of the self-employed is discussed in Blanchflower and Oswald (1992).
9 Positive coefficients in the ordered probit regressions are associated with a higher estimated
probability that the individual reports job satisfaction of greater than level i (see Greene, 1993). The
ordered probit procedure chooses estimates b to maximise ~ln(p,.), where Pi is the estimated
probability of the observed response and the summation is over all of the observations in the data set.
The probability of observing level i is Pi = Pr(/2i- I < x ' b + u <_ i~i), where u is assumed to be
normally distributed. The b coefficients are estimated by the procedure, as are the thresholds, /21, /22,
.... /2/- 1, where I is the number of categories of the ordered dependent variable. /20 is taken to be
-oo and /21 + oo; the probabilities thus sum to t. Say that an individual j has characteristics xj such
that x'jb = - 0 . 7 . Then the estimated probability that j reports overall job satisfaction of 7, using the
estimates of the/2's from Table 2, is Pr(-0.411 < - 0 . 7 + u < + ~ ) - - Pr(0.289 < u)--- 1 - F(0.289)=
38.6% (where F is the cumulative normal distribution). Similarly, the estimated probability that j
responds '6' is P r ( - 1.161 < - 0 , 7 + u_< - 0 . 4 1 1 ) = 1 - F ( - 0 . 4 6 1 ) - ( 1 - F(0.289)) = F ( 0 , 2 8 9 ) -
F ( - 0 . 4 6 1 ) = 29.1%. The estimated probability that j responds '1' is F ( - 2 . 1 7 5 ) = 1.5%.
10 All of these correlations are discussed in more detail in Clark (1996c).
A.E. Clark/Labour Economics 4 (1997) 341-372 349

ship between age and all measures of job satisfaction tl Good health has a
consistent, large and well-defined positive effect on job satisfaction, but higher
levels of education are associated with less-satisfied workers ~2. Both marital
status dummies and ethnic background dummies are only occasionally significant.
Renters are more satisfied than homeowners, perhaps because it is easier for them
to leave jobs that they dislike.
With respect to the job, longer hours of work are associated with lower
satisfaction (especially for satisfaction with pay), as is the time spent travelling to
work. Income has no effect on overall job satisfaction but is strongly positively
correlated with satisfaction with pay and negatively correlated with relations with
manager, this latter perhaps reflecting professional frictions at more elevated
positions. Those with managerial responsibilities are more satisfied, as are those
who are not members of trade unions or who are in permanent positions.
Promotion opportunities are significantly positively correlated with all measures of
job satisfaction. Longer job tenure increases satisfaction with job security and the
use of initiative, although at some cost in terms of relations with superiors. Lastly,
there is a job satisfaction premium from working in a small establishment (see
Idson, 1990). This satisfaction boost is sharpest for the intrinsic measures of job
satisfaction, relation with supervisors, use of initiative and the work itself;
establishment size is not correlated with the extrinsic measures of satisfaction, pay
and hours of work. This is consistent with the interpretation of larger establish-
ments' higher wages (Brown and Medoff, 1989) as a compensating differential to
make up for the less interesting and challenging work that is carried out in them
(Schmidt and Zimmerman, 1991).
The size of the estimated effects on job satisfaction can be illustrated by
considering the predicted probabilities that a certain 'baseline' individual reports
different levels of job satisfaction. These baseline individuals are aged 40, in
excellent health, with medium education, married, house owners, non-union,
working 35 hours per week, in an establishment of size 25-199, in a
managerial/professional occupation, in the metal goods, engineering or vehicles
industries (SIC4). Using the results from column 1 of Tables 2 and 3, a woman
with these characteristics has a 40% probability of reporting job satisfaction of 7,
whereas an identical man has only a 31% probability. From this baseline, large

11 The estimated minima of these convex relationships lie between the ages of 24 for satisfaction with
the work itself to 40 for satisfaction with job security. See Clark et al. (1996) for a more detailed
discussion of age and well-being using the same dataset.
12 The education result is one of the more surprising findings, as it is well-known that the
better-educated have better jobs. One explanation is that education does bring rewards, but also raises
expectations (Hagenaars, I986, Ross and Reskin, 1992), leading to greater disappointment and
dissatisfaction. Another is that higher-educated workers are more likely to suffer from educational
mismatch (Sloane et al., 1995). Unfortunately, it is not possible to test this directly using the BHPS
dataset; however, Groot and Maassen van den Brink (1996) find no independent effect of over-educa-
tion on reported job satisfaction in a sample of Dutch workers.
350 A.E. Clark~Labour Economics 4 (1997) 341-372

Table 2
Basic satisfaction regressions
Overall Promotion Pay Relations
Male -0.240 -0. ! 23 -0.274 -0.142
(0.038) (0.039) (0.041) (0.038)
Age -0.031 -0.033 -0.042 -0.021
(0.008) (0,009) (0,009) (0.010)
Age-squared/1000 0.53 t 0.555 0.623 0.439
(0.101) (0.110) (0.107) (0,126)
Health'. Excellent 0.376 0,201 0.214 0.341
(0.046) (0.048) (0.047) (0.049)
Health: Good 0.147 0.133 0.143 0.144
(0.042) (0.045) (0.044) (0.046)
Education: Higher -0.328 -0.169 -0.123
(0.051) (0.052) (0.049)
Education: A / O / N u r s i n g -0.225 -0.191 -0.067
(0.041) (0.043) (0.043)
Married 0. t 19
(0.053)
Cohabiting -0.073
(0.067)
Widowed 0.149
(0.167)
Divorced 0.020
(0.098)
Separated 0.079
(0.140)
Renter 0.115 0.090 0.087
(0,040) (0.041) (0.043)
Log income 0.453 -0.145
(0,039) (0.028)
Log hours -0.135 -0.138 -0.574
(0.030) (0.032) (0.042)
Union -0.118 -0.133 -0.083 -0.124
(0.038) (0.037) (0.038) (0.038)
Promotion opportunities 0.259 0.797 0.191 0.170
(0.035) (0.037) (0.036) (0.037)
Job tenure (months) -0.001
(0.000)
Establishment size: t-24 0.215 0.184 0.252
(0.044) (0.044) (0.046)
Establishment size: 25-199 0.070 -0.011 0,062
(0.039) (0.040) (0.041)
Region dummies ~ tJ
Industry dummies u," u,~
Occupation dummies u*" ~," ~,~
Mu (1) -2.875 - 1.581 - ! .503 -2.590
Mu (2) -2.672 -1.389 -1.288 -2.382
Mu (3) -2.321 -1.163 -0.920 -2.158
Mu (4) - 1.746 -0.336 -0.352 - 1.604
A.E. Clark~Labour Economics 4 (1997) 341-372 351

Table 2 (continued)
Overall Promotion Pay Relations
Mu (5) -1.161 0.002 0.119 -1.231
Mu (6) -0.411 0.310 0.571 -0.703
N 4743 4248 4315 4320
Log-likelihood -7418.97 -7272.24 -7858.10 -6504.41
Log-likelihood at zero -7694.89 -7593.25 -8076.48 -6687.13

Standard errors in parentheses.

effects result from age (increasing the individual's age from 40 to 60 changes
these probabilities to 57 and 48%, respectively) and from having the worst level of
health (27 and 19%). From the estimated coefficients, substantial effects on
predicted probabilities are also associated with education, promotion opportunities
and firm size. Renter, union membership and hours of work all have rather smaller
effects: an increase in hours to 50 per week only reduces the predicted probability
of reporting overall job satisfaction of 7 to 38% and 30% for women and men,
respectively.
The distribution of these objective job and individual characteristics does help
us to understand which types of worker are satisfied, but not why women are more
satisfied than men. The next section considers whether a subjective variable, work
values, might explain the gender satisfaction differential.

4. Work values
The second of Section 2's explanations invoked the different values which men
and women have with respect to work. In the BHPS dataset, all respondents in the
labour force are asked to indicate the first and second most important aspects of a
job. Respondents choose from a list of seven specific job attributes, the same as
those for the specific job satisfaction questions described above, plus an eighth
catch-all category, 'something else'. Work values differ significantly by gender, as
shown in Table 4. Men rank promotion prospects, pay and job security more
highly than do women, but women are significantly more likely to mention good
relations with managers, the actual work itself and the hours of work. The gender
difference in the percentage citing each of the eight aspects as the most important
is significant at the 5% level; for six of the eight categories this difference is
significant at the 0.1% level. These results are consistent with the view that men
are more concerned with the extrinsic aspects of work (such as pay and promo-
tion) whereas women are more likely to value the intrinsic returns to work 13

~3 Similar findings are reported for the US by Neil and Snizek (1987); see also Garai and Scheinfeld
(1968) and Moir and Jessel (1989). The link between values and job satisfaction has been discussed by
de Vaus and McAllister (1991), Gruenberg (1980), Hakim (1991), Janson and Martin (1982) and
Morris and Villemez (1992).
352 A.E. Clark / Labour Economics 4 (1997) 341-372

Table 3
Basic satisfaction regressions
Job security Initiative Work itself Hours
Male -0.077 (0.038) -0.053 (0.039) -0.218 (0.038) -0.107 (0.041)

Age -0.063 (0.008) -0.026 (0.010) -0.021 (0.008) -0.032 (0.009)

Age-squared/1000 0,790 (0.104) 0.424 (0.119) 0.477 (0.104) 0.536 (0.113)

Health: Excellent 0.130 (0.047) 0.190 (0.048) 0.286 (0.047) 0.281 (0.048)

Health: Good 0.077 (0.044) 0.045 (0.045) 0.113 (0.043) 0.167 (0.045)

Education: Higher -0.137 (0.046) -0.247 (0.054) -0.332 (0.052) -0.195 (0.054)

Education: A/O/Nursing -0.083 (0.041) -0.107 (0.043) -0.196 (0.042) -0.127 (0.044)

Married 0.226 (0.053)

Cohabiting 0.077 (0,066)

Widowed 0.560 (0. L65)

Divorced 0.372 (0.097)

Separated 0.333 (0.137)

Black 0.061 (0.145)

Asian subcontinent 0.358 (0.139)

Renter 0.082 (0,041) 0.140 (0.043) 0.089 (0.041) 0.119 (0,043)

Log income -0.116 (0.040)

Log hours -0.089 (0.031) -0.143 (0.043)

Travel time/100 -0.159 (0.075) -0.245 (0.078)

Manager 0.220 (0.041) 0.113 (0.040)

Union -0,219 (0.039) -0.197 (0.040) -0.114 (0.038)

Temporary worker -0.792 (0.064) -0.190 (0.061) -0.219 (0.064)

Promotion opportunities 0.276 (0.035) 0.135 (0.037) 0.108 (0.035) 0.078 (0.036)

Job tenure (months) 0.001 (0.000) 0.001(0.000)

Establishment size: 1-24 0.175 (0.045) 0.325 (0.045) 0.277 (0.044)


A.E. Clark/Labour Economics 4 (1997) 341-372 353

Table 3 (continued)
Job security Initiative Work itself Hours
Establishment size: 25-199 0.046 (0,040) 0,099 (0,041) 0.083 (0.040)

Region dummies u*' u,~ ua u*'


Industry dummies u,~ u*' u,~ tl
Occupation dummies ~ ~ u"
Mu (1) -3.196 - 1.771 - 1.674 -3.530
Mu (2) -2.994 - 1.583 - 1.520 -3.279
Mu (3) -2.737 -1.334 -1.228 -2,932
Mu (4) -2.207 -0.820 -0.601 -2.295
Mu (5) -1,867 -0.396 -0.133 -1.908
Mu (6) - 1.426 0.189 0.477 - 1.455
N 4647 4783 4814 4388
Log-likelihood -7583.78 -6689,01 -7149.95 -7131.78
Log-likelihood at zero -7871.90 -6931.34 -7426.38 -7361.08

Standard errors in parentheses.

Table 5 presents regression results including work values variables. The


excluded categories are 'something else' and 'initiative', leaving six dummy
variables for each of the first and second most important aspects of a job. For this
and all subsequent statistical work only satisfaction with pay, satisfaction with
work itself and overall job satisfaction will be analysed. These three measures are
broad in scope in that the first captures satisfaction with an extrinsic aspect of the
job, the second with an intrinsic reward and the third is a summary measure of

Table 4
The most important aspects of a job
Percentage saying most important Percentage saying second most important
all men women t-statistic all men women t-statistic
Promotion 2,7 3.5 2.0 3,14 5.0 5.8 4.3 2.38
Pay 16.3 19.2 13.4 5.54 24,7 27.6 21.9 4.69
Relations at work 7.6 4.6 10.4 7.76 14,1 10,5 17.6 7.26
Job security 30.0 35.7 24,5 8.66 17.6 20.2 15.1 4.78
Initiative 8.4 9.4 7.5 2.35 13.5 12.5 14.5 2.07
Work itself 29,2 23.9 34.5 8.18 16.9 17.4 16.3 1.04
Hours 3.6 1.2 6.0 9.09 6.6 4.6 8.5 5.46
Something else 2.1 2.5 i .7 1.96 1.6 1.3 1.9 1.42
354 A.E. Clark/Labour Economics 4 (1997) 341-372

Table 5
Job satisfaction regressions with work values
Overall satisfaction Pay satisfaction Satisfaction with
work itself
Male - 0 . 2 0 4 (0.039) - 0 . 2 2 3 (0,041) - 0 . 1 7 9 (0.039)
Age - 0.047 (0.009) - 0.048 (0.008) - 0.022 (0.008)
Age-squared/1000 0.665 (0.112) 0.671 (0.107) 0.475 (0.106)
Health: Excellent 0.384 (0.046) 0.194 (0.047) 0.273 (0.047)
Health: Good 0.150 (0.043) 0. ! 29 (0.044) 0.109 (0.043)
Education: Higher - 0 . 3 0 4 (0.052) - 0 . 3 2 8 (0.053)
Education: A / O / N u r s i n g - 0.189 (0.042) - 0.183 (0.043)
Married 0.170 (0.051)
Cohabiting 0.049 (0.064)
Widowed 0.348 (0.152)
Divorced 0.083 (0.090)
Separated 0.179 (0.130)
Renter 0.096 (0,040) 0,084 (0,041)
Log income 0.460 (0.039)
Log hours - 0.119 (0,031) - 0.585 (0.042)
Manager 0.120 (0.040)
Union - 0.113 (0.037) - 0.095 (0.038)
Promotion opportunities 0.255 (0,035) 0.188 (0.035) 0.108 (0.035)
Establishment size: 1-24 0.193 (0,044) 0.262 (0,044)
Establishment size: 25-199 0.068 (0.039) 0.083 (0.040)
Region dummies ~ tJ
Industry dummies ~ ~
Occupation dummies ~ ~ ~'
Work Values: First Mention
Promotion prospects - 0 . 2 7 6 (0.104) - 0 . 3 2 0 (0.105) - 0 . 1 8 2 (0.105)
Total pay -0.341 (0.064) - 0 . 1 9 0 (0.065) - 0 . 2 9 9 (0.064)
Relations at work 0.174 (0.079) 0.096 (0.080) 0.131 (0.081)
Job security 0.136 (0.059) 0.032 (0.060) 0.045 (0.060)
Actual work itself - 0 . 0 0 6 (0.060) - 0 , 0 0 7 (0.061) 0.096 (0.061)
Hours 0.121 (0.104) 0.210 (0.103) 0,102 (0.104)
Work Values: Second Mention
Promotion prospects - 0 . 1 6 2 (0.081) - 0 . 3 1 2 (0.081) -0.111 (0.082)
Total pay - 0 . 1 7 6 (0.053) - 0 . 2 1 7 (0.054) - 0 . 0 7 3 (0.054)
Relations at work 0,176 (0.061) 0.055 (0.061) 0.112 (0.062)
Job security 0.024 (0,058) - 0.150 (0.059) 0,065 (0.059)
Actual work itself 0.024 (0.060) - 0 . 0 2 5 (0.061) 0.102 (0,061)
Hours 0,019 (0.076) - 0 . 0 7 2 (0.076) - 0 . 0 1 4 (0,077)
Mu ( 1) - 3,036 (0.223) - 1.712 (0.209) - 1.694 (0,193)
Mu (2) - 2.834 (0.222) - 1,492 (0.209) - 1.540 (0.193)
Mu (3) - 2 . 4 7 9 (0,221) - 1.123 (0.209) - 1.248 (0.t92)
Mu (4) - 1.897 (0.221) - 0 . 5 5 3 (0.208) - 0 . 6 1 6 (0.191)
Mu (5) - 1,301 (0,220) - 0 . 0 7 7 (0.208) - 0 . 1 4 4 (0.191)
Mu (6) - 0 , 5 4 0 (0,220) 0.380 (0.208) 0.473 (0.191)
N 4718 4416 4794
Log-likelihood - 7324.78 - 8005.20 - 7086,94
Log-likelihood at zero - 7657.28 - 8267,01 - 7399.34
A.E. Clark/Labour Economics 4 (1997) 341-372 355

Table 5 (continued)
Overall satisfaction Pay satisfaction Satisfaction with
work itself
Exclusion of work values variables (×2(6))
tst mention: 105.3 39.9 71.2
2nd mention: 49.2 45.4 20.5

Standard errors in parentheses.

satisfaction with all aspects of work 14. Also, from Tables 2 and 3, these are the
three measures for which the gender differential is the most significant.
The results show that both sets of work values variables (entered as sets of six
(1, 0) dummies) are highly statistically significant: workers' values are important
predictors of worker well-being. Specifically, workers who say that pay or
promotion is important report significantly lower job satisfaction, whereas an
emphasis on relations at work is associated with higher job satisfaction. However,
the estimate on the 'male' dummy variable shows that a female worker with the
same personal characteristics (age, education, marital status etc.), doing the same
job (pay, hours, occupation etc.) and with the same work values remains statisti-
cally significantly more likely to report a higher job satisfaction score than her
male counterpart.

5. Self-selection into the labour market and job satisfaction

The third explanation of women's higher reported levels of job satisfaction


relies not on the specification or measurement of the variables influencing
well-being, but rather on men's and women's different participation rates. The
(unobserved) 'true' distribution of well-being at work may be identical across
gender (once other explanatory variables are controlled for) but, if dissatisfied
women are less likely to be in employment than dissatisfied men, the observed
distribution of job satisfaction will be biased for the usual reasons.
In the BHPS dataset, 67% of women aged 16-64 were active in the labour
force, compared to 86% of men. However, because men are more likely than
women to be self-employed or unemployed, the difference in employment rates,
which is what counts for this analysis, is less striking: 58.6% of working age
women as against 63% of men. The underlying assumption is that potential job
satisfaction is related to the probability of being an employee; if there is no

14 Overall job satisfaction seems to be a good summary measure of the specific job satisfaction
questions, being strongly individually correlated with all seven of them. In addition, an ordered probit
regression of overall job satisfaction on 42 dummy variables calculated from the seven specific job
satisfaction measures produced estimates significant at the 1% level or better for all 42 dummies, with
38 estimates having t-statistics of over 4.
356 A.E. Clark~Labour Economics 4 (1997) 341-372

Table 6
OLS job satisfaction regressions with selection correction
First principal first principal
component of the component of
seven specific job overall satis-
satisfaction faction, pay
questions (FPC7) satisfaction, and
satisfaction with
the work itself (FPC3)
Male - O. 156 (0.037) - 0.202 (0.036)
Age -o.o41 (O.OLO) - 0.027 (0.009)
Age-squared/1000 0.637 (0.126) 0.436 (t3.114)
Health: Excellent 0.336 (0.044) 0.330 (0.042)
Health: Good 0.213 (0.041) O. 182 (0.039)
Education: Higher - 0.207 (0.049) - O. 198 (0.047)
Education: A / O / N u r s i n g - 0 . 1 3 5 (0.040) - 0 . 1 3 1 (0.038)
Married O. 129 (0.048) O. 128 (0.046)
Cohabiting 0.051 (0.060)
- - 0.037 (0.059)
Widowed 0.032 (0.127) 0.003 (0.121)
Divorced 0.088 (0.086) 0.024 (0.082)
Separated 0.217 (0.153) 0.230 (0.142)
Renter 0.094 (0.039) 0.052 (0.037)
Log hours - 0.097 (0.032) - 0.081 (0.029)
Union - 0 . i 14 (0.035) - 0.090 (0.034)
Promotion opportunities 0.450 (0.033) 0.264 (0.032)
Establishment size: 1-24 0.239 (0.041) O. 172 (0.040)
Establishment size: 25-199 0.089 (0.037) 0.069 (0.036)
Industry dummies /J l,J
Occupation dummies t.J

Work values: first mention


Promotion prospects - O. 187 (0.099) - 0.333 (0.096)
Total pay - 0.288 (0.061) -0.318 (0.058)
Relations at work O. 154 (0.076) 0.089 (0.072)
Job security 0.062 (0.057) 0.062 (0.054)
Actual work itself - 0.060 (0.057) - 0.0o8 (0.055)
Hours 0.059 (0.100) 0.087 (0.093)

Work values: second mention


Promotion prospects - 0.239 (0.076) - 0 . 2 5 5 (0.074)
Total pay - 0 . 1 8 6 (0.051) - 0 . 1 5 8 (0.049)
Relations at work 0.066 (0.058) 0.100 (0.055)
Job security - 0.083 (0.055) - 0.o40 (0.053)
Actual work itself - 0.052 (0.057) 0.023 (0.054)
Hours - 0.093 (0.075) - 0.062 (0.070)
Constant 0.568 (0.253) 0.401 (0.225)
Lambda 0.027 (0.055) 0.066 (0.048)
Number of observations 4088 4618

Standard errors in parentheses.


A.E. Clark/Labour Economics 4 (1997) 341-372 357

relation then even non-random participation will yield a random sample of job
satisfaction responses.
The technique employed is a full maximum likelihood estimation of the
Heckman selection model. Two composite measures of job satisfaction are used as
dependent variables in the linear regression part of the procedure: the first
principal component of the seven specific job satisfaction equations discussed in
Section 3 (henceforth referred to as FPC7) and the flu'st principal component of the
three job satisfaction measures analysed in Table 5 (FPC3). The probit model for
being an employee included all of the non-job variables in Table 5's regression
specification, plus a number of other, mainly household, variables which help to
determine the probability of employment, but are assumed not to influence
reported levels of job satisfaction. These latter are variables which identify the
selection model. They include spouse's pay, spouse's hours of work, number of
children in the household, the household division of tasks such as shopping or
cooking, and the provision of care for others (all of these interacted with gender),
as well as the income of others in the household and the individual's own
unearned income.
The estimated employment probit equations are very similar for the two
satisfaction measures analysed: that for FPC7 is presented in Appendix C. The
sample used to estimate this probit consists of all BHPS respondents under the age
of retirement (65 for men, 60 for women). The results show that the estimated
probability of employment is higher for males, whites, the middle-aged, those in
good health, those with higher levels of education and house buyers. Individuals in
households with greater numbers of others employed, with higher unearned
income and with more or younger children are less likely to be in employment.
The female married are the least likely to be employed, while the male married are
the most likely to be in employment. The job satisfaction results are presented in
Table 6; the specification is the same as in Table 5. The estimates on the 'male'
variable are negative and significant for both satisfaction measures, being only
little changed from the estimates resulting from an OLS regression without
correction for sample selection. This conclusion has been checked for robustness
by the use of a number of different specifications of the employment probit, by the
use of a simpler two-step estimator and by the use of the three measures of job
satisfaction in Table 5 considered as cardinal variables. Whatever the measure and
whatever the specification, in this dataset sample selection does not account for
women's higher job satisfaction.

6. Expectations

The last explanation proposed in Section 2 rests on the presence of relative


terms in the well-being function, which may refer to an individual's expectations
about their job: women will be more satisfied than a man with the same objective
358 A.E. Clark~Labour Economics 4 (1997) 341-372

Table 7
Job satisfaction regressions with comparison income
Overall satisfaction Pay satisfaction Satisfaction with
work itself
Male - 0.176 (0.044) 0.187 (0.046)
- - 0.155 (0.047)
Age - 0 . 0 4 0 (0.011) - 0.042 (0.010) - 0 . 0 1 3 (0.010)
Age-squared/1000 0.561 (0.130) 0.57t (0.126) 0.339 (0.124)
Health: Excellent 0.434 (0.050) 0.265 (0.049) 0.305 (0.051)
Health: Good O. 176 (0.047) 0.153 (0.046) 0.137 (0.047)
Education: Higher - 0.263 (0.056) - 0.344 (0.059)
Education: A / O / N u r s i n g - 0 . 1 5 1 (0.046) - 0.213 (0.047)
Married O. 197 (0.055) 0.131 (0.054)
Cohabiting 0.054 (0.068) - 0.056 (0.067)
Widowed 0.282 (0.167) 0.097 (0.161)
Divorced 0.055 (0.098) - 0.013 (0.096)
Separated 0.088 (0.144) 0.061 (0.I40)
Renter 0.118 (0.044) O. 106 (0.044)
Log income O. 147 (0.050) 0.629 (0.049) 0.099 (0.047)
Log predicted income - 0.237 (0.066) - 0.405 (0.063) 0.185 (0.064)
-

Log hours - 0.128 (0.049) - 0.444 (0.049)


Manager 0.129 (0.042) 0.170 (0.045)
Union - O. 106 (0.040) - 0.087 (0.041)
Temporary worker - 0 . 1 5 7 (0.071)
Second job - 0.124 (0.058)
Promotion opportunities 0.245 (0.038) 0.225 (0.037) 0.103 (0.039)
Establishment size: 1-24 O. 133 (0.049) O. 198 (0.050)
Establishment size: 25-199 0.067 (0.042) 0.088 (o.o43)
Industry dummies tJ
Occupation dummies Ua

Work values: first mention


Promotion prospects -0.241 (0.110) - 0 . 3 5 8 (0.110) - 0 . 1 6 6 (0.111)
Total pay - 0.347 (0.070) - 0.214 (0.069) - 0.296 (0,070)
Relations at work 0,214 (0.088) 0.042 (0.085) O. 125 (0.090)
Job security 0.163 (0.065) 0.034 (0.064) 0.089 (0.066)
Actual work itself 0.046 (0.065) - 0.005 (0.065) 0.136 (0.066)
Hours 0.119(0.117) 0.063 (0.113) 0.086 (0.118)

Work values: second mention


Promotion prospects - 0.192 (0.086) - 0.346 (0.085) - 0.102 (0.087)
Total pay - 0.163 (0.058) - 0.228 (0.057) - 0 . 0 5 6 (0.059)
Relations at work 0.190 (0.066) 0.028 (0.064) 0.163 (0.068)
Job security 0.042 (0.064) - O. 129 (0.062) 0.104 (0.065)
Actual work itself 0.027 (0.065) - 0.032 (0.064) O. 115 (0.066)
Hours o.ol 1 (0.083) - 0.085 (0.081) - 0.024 (0.084)
Mu (1) - 3.544 (0.295) - 2.289 (0.257) - 1.971 (0.285)
Mu (2) - 3.338 (0.295) - 2.075 (0.257) - 1.8o4 (0.285)
Mu (3) - 2.967 (0.294) - 1.698 (0.257) - 1.494 ( 0 . 2 8 4 )
Mu (4) - 2.379 (0.293) - 1.115 (0.256) - 0.875 (0.284)
A.E. Clark/Labour Economics 4 (1997) 341-372 359

Table 7 (continued)
Overall satisfaction Pay satisfaction Satisfaction with work itself

M u (5) - 1.780 (0.293) - 0 . 6 2 5 (0.256) - 0.382 (0.284)


M u (6) - 0 . 9 9 9 (0.292) - 0 . 1 5 6 (0.256) 0.235 (0.284)
N 3972 3982 4023
Log-likelihood - 6196.85 - 7201.81 - 6049.04
Log-likelihood at zero - 6485.40 - 7452.54 - 6301.90

Standard errors in parentheses.

characteristics and work values if women expect less than men from their job.
Empirical evidence on women's lower expectations is slight. Some small-sample
results are presented in Hodson (1985) and Major (1987). Sloane and Williams
(1995), using the British SCELI dataset, find that 63% of men feel that they
receive lower pay than they deserve, against only 48% of women. More generally,
Waldman (1994) comments on the systematic over-prediction by men of their
potential performance at a number of tasks, to a far greater extent than by women.
A first test of this explanation consists in including an explicit comparison
income variable in the job satisfaction regressions. Following Clark and Oswald
(1996), a predicted wage is computed for every worker in the sample, using a wide
range of individual and job characteristics, including a dummy variable for sex. As
is usually found, this gender variable is an important predictor of earnings: ceteris
paribus, women receive lower wages than men. The estimates from this income
equation are reported in Appendix D. The results in Table 7 show that, as
expected, predicted income is negatively significantly correlated with all three
measures of job satisfaction, but that the estimated 'male' coefficient in the job
satisfaction regressions, although somewhat reduced in size, remains negative and
significant. The use of a measure of y * from separate income regressions by
gender, to reflect the possibility that men and women are rewarded differently for
their individual and job characteristics, does not alter this conclusion.
Table 8 shows the evolution of the female-male job satisfaction differential
with various sets of controls. For each satisfaction measure, these estimates come
from regressions run on the same sample of employees, those with reported
information on all of the control variables.
The standard set of controls reduces the overall job satisfaction gender differen-
tial by about a quarter, but actually raises the differentials for satisfaction with pay
and satisfaction with the work itself, Work values are important predictors of
satisfaction and their inclusion reduces the gender differential. With the introduc-
tion of estimated comparison income levels, in addition to the previous regressors,
the gender differential is half of its raw data level for overall job satisfaction, 80%
of its original level for satisfaction with work itself, but only slightly reduced for
satisfaction with pay.
360 A.E. Clark/Labour Economics 4 (1997) 341-372

Table 8
Coefficient on 'male'
Controls Overall job Satisfaction Satisfaction with
satisfaction with pay the work itself
No controls -0.325(0.034) - 0.200 (0.033) - 0.192 (0.034)
Standard (Tables 2 and 3) - 0 . 2 5 5 (0.042) - 0.308 (0.042) - 0.228 (0.041)
Standard + work values (Table 5) - 0 . 2 2 6 (0.043) - 0.273 (0.043) - 0.191 (0.042)
Standard + work values + predicted - 0 . 1 7 6 (0.044) - O. 187 (0.046) - 0.155 (0.047)
income (Table 7)
Standard + work values + predicted - 0.167 (0.044) -0.192(0.045) - 0.146 (0.046)
income (by sex)

As discussed in Section 2, the process of comparison likely takes place over a


wide range of job attributes. A more indirect approach consists in finding variables
that might realistically be correlated with an individual's job expectations. It is
likely that at least part of these expectations are formulated quite early in life by
education and upbringing (Crosby, 1982; Hodson, 1985; Loscocco and Spitze,
1991; Miller, 1980; Ross and Reskin, 1992). A first test is then to investigate the
relationship between the female-male job satisfaction differential and respon-
dents' age and education. The supposition is that any gender difference in
expectations should be smaller for younger workers and for better-educated
workers. The first part of Table 9 reports the results of regressions with age and
education interactions by gender: these show that the female-male satisfaction
differential does largely rise with age, with the youngest female workers reporting
the same overall job satisfaction and satisfaction with work itself as their male
counterparts, and lower satisfaction with pay. There is little difference in the
satisfaction scores by sex for workers over the age of 60, but this may come from
the self-selection of older dissatisfied workers of both sexes out of the labour
force. The education results are also consistent with the hypothesis: there is no
gender difference in the job satisfaction reported by workers with higher levels of
education, whereas the gender satisfaction differential is significant at the 0.1%
level for workers in the other two education groups.
The third part of Table 9 tests the influence of another potential correlate with
job expectations: mother's labour force status. The BHPS dataset includes infor-
mation on parent's labour force status when the respondent was aged 14. Regres-
sions with an interaction term between gender and three dummies representing
mother's labour force status show that highly significant gender job satisfaction
differentials are obtained for respondents whose mother was not at work or who
worked in a non-professional position; the satisfaction differential for those with a
professional mother is smaller and mostly insignificant.
If low expectations are the source of women's higher job satisfaction, it is of
A.E. Clark/Labour Economics 4 (1997) 341-372 361

Table 9
Gender interactions in job satisfaction regressions

Estimated coefficients on the 'male' variable in job satisfaction regressions

overall pay satisfaction


satisfaction satisfaction with work
itself

Age
16-19 0.033 0.142 -0.015
20-29 -0.185 * * -0.123 -0.203 **
30-39 -0.256" * * -0.372 * ** -0.219 * *
40-49 - 0.189 * * - 0.265 * * * - 0.057
50-59 -0.360* ** -0.457* ** -0.311" * *
60 + - 0.170 - 0.301 - 0.210

Education
Degree - 0.029 - 0.046 0.028
A/O/Nursing - 0,294" * ~ - 0.224" * * - 0.233 * *
Other - 0.275 * * * - 0.432 * * * - 0.301 * * "

Working mother
No -0.253* * * -0.274' * * -0.199' * *
Non-professional - 0.213 * * * -0.161 * * -0.209 * * '
Professional - 0. ! 6 4 - 0.243 * - 0.160

Occupation
Managers and administrators - 0.141 - 0.126 0.038
Professional - 0.062 0.170 - 0.058
Associate professional and technical -0.104 -0.073 -0,128
Clerical and secretarial - 0.421 * * * - 0.356 * * * - 0.382 * * *
Craft and related - 0.392 * * - 0.333 * - 0,357'
Personal and protective service -0.283 * * -0.093 -0.531 * **
Sales - 0.067 -0.317 * * 0.172
Plant and machine operatives 0.026 -0.478 * ** 0.024
Other - 0.311 * * - 0.540 * * ' - 0,193

Sex mix at place of work


Mainly or almost exclusively men - 0.060 - 0.334 * * * - 0,053
Equal men and women -0.245* * * -0.296* * * -0.222* *
Mainly or almost exclusively women - 0.328* * * - 0.156 - 0.387 * *

Significant at the 5% level.


* ' Significant at the 1% level.
.... S i g n i f i c a n t at t h e 0 . 1 % l e v e l .

interest to ask whether these expectations might be changed by subsequent labour


market experience. Specifically, exposure to 'good' jobs might be expected to
revise these expectations upwards. This proposition is investigated by considering
interactions between gender and occupation: if this hypothesis is true then the
gender differential should be lower in professional occupations. The results in
362 A.E. Clark/Labour Economics 4 (1997) 341-372

Table 9 show that, for all three satisfaction measures, the gender differential is
insignificant for workers in professional classifications; in contrast, there is a
significant gender differential in at least one of the job satisfaction measures for
workers in all of the non-professional occupations.
As well as education and professional experience, workers' expectations about
what kind of job they should receive may be influenced by observation of others at
the place of work. The last part of Table 9 uses information on sex mix at the
respondent's place of work. This has been combined into three dummies for
mainly male, mixed and mainly female workplaces. The prior belief is that women
in male-dominated workplaces may expect more, reducing the gender satisfaction
differential (Loscocco and Spitze, I991). For overall job satisfaction and satisfac-
tion with the work itself, this is borne out: there is no differential for those in male
workplaces, a larger and significant differential for those in mixed workplaces,
with the largest differential pertaining to those in female workplaces. However, for
satisfaction with pay, the relationship is the opposite. It is true that both men and
women earn less in female-dominated workplaces, but these results are holding
wages constant. The behaviour of the pay satisfaction gender differential remains
an anomaly in the light of the other results reported above.

7. The determinants of male and female job satisfaction

The final part of this paper looks at the separate determinants of male and
female job satisfaction. The results of separate overall job satisfaction regressions
by gender are reported in Table 10.
Age and health have comparable effects on men's and women's overall job
satisfaction. However, the negative effect of higher education on satisfaction is
more marked for women than for men (see Miller, 1980). Marital status, renter,
hours of work and union and managerial status are significant determinants of
overall job satisfaction only for women. There are a number of differences in the
effects of the work values variables by gender. Choosing pay or promotion as the
first most important aspect of a job has a much larger negative effect on men's job
satisfaction than on women's; conversely, the choice of job security and relations
at work is associated with significantly higher job satisfaction for women but has
no effect for men. Women who say that hours of work is the first most important
aspect of a job report significantly higher levels of job satisfaction, while the effect
for men is negative (but insignificant). Marriage has a strong positive effect on
women's job satisfaction but not on men's, which is consistent with a selection
interpretation of the gender job satisfaction differential. However, the positive
coefficients on divorced, separated or widowed run contrary to the simple partici-
pation effect (all of these groups are more likely to be employed, according to the
probit results in Appendix C).
A.E. Clark/Labour Economics 4 (1997) 341-372 363

Table 10
Overall job satisfaction regressions: Men and women
Women Men
Age - 0.044 (0.014) - 0.060 (0.011)
Age-squared/1000 0,619 (0.171) 0,837 (0.134)
Health: excellent 0.419 (0.066) 0.309 (0.063)
Health: good 0.214 (0.060) 0.068 (0.060)
Education: higher - 0.387 (0.079) - 0.211 (0.068)
Education: A / O / n u r s i n g - 0 . 1 5 2 (0,060) - 0 . 2 0 6 (0,058)
Married 0.264 (0.076)
Cohabiting 0.091 (0.090)
Widowed 0.442 (0.180)
Divorced 0.209 (0.120)
Separated 0.246 (0. t 57)
Renter 0.136 (0.058)
Log hours - 0.118 (0.038)
Manager 0.147 (0.058)
Union - 0.104 (0.052)
Second job - 0,153 (0.077)
Promotion opportunities 0.198 (0.051) 0.296 (0,047)
Establishment size: 1 - 2 4 0.169 (0.061) 0.235 (0.060)
Establishment size: 2 5 - 1 9 9 - 0.020 (0.059) 0.163 (0.052)
Industry dummies tJ
Occupation dummies u'~ u,"

Work values: first mention


Promotion prospects - 0.141 (0.172) - 0.350 (0.130)
Total pay - 0 . 2 4 4 (0,097) - 0 , 4 2 8 (0.083)
Relations at work 0.277 (0.107) 0.107 (0,121)
Job security 0.241 (0.090) 0,021 (0~077)
Actual work itself 0.089 (0.088) - 0 , 0 9 1 (0.080)
Hours 0.257 (0.126) - 0.292 (0.206)

Work values: second mention


Promotion prospects - 0.215 (0.120) - 0.018 (0.108)
Total pay - 0 . 2 0 8 (0.077) - 0 . 1 1 9 (0.073)
Relations at work 0.095 (0.081) 0,274 (0,090)
Job security - 0.065 (0.084) 0,128 (0.080)
Actual work itself - 0 , 0 3 5 (0,086) 0.111 (0.082)
Hours 0.029 (0.101) - 0,038 (0.115)
Mu (1) - 2,703 (0.320) - 2.648 (0.249)
Mu (2) - 2.498 (0,319) - 2.436 (0,248)
Mu (3) - 2.174 (0.318) - 2.048 (0.247)
Mu (4) - 1.592 (0.317) - 1,462 (0.246)
Mu (5) - 1.028 (0.316) - 0 . 8 4 6 (0.245)
Mu (6) - 0.300 (0.316) - 0.047 (0,245)
N 2354 2473
Log-likelihood - 3456.33 - 4011.13
Log-likelihood at zero - 3613,46 - 4162.91

Standard errors in parentheses.


364 A.E. Clark/Labour Economics 4 (1997) 341-372

Table 11
Predicted probabilities
Women Men
P(7) (%) P(6 or 7) (%) P(7) (%) P(6 or 7) (%)
Baseline 35.8 64.2 18.7 46.4
Baseline + higher education 27.4 55.1 18.6 46.2
Baseline + never married 26.4 53.9 18.7 46.4
Baseline + hours = 20 p.w. 38.2 66.6 18.7 46.4
Baseline + no promotion opps 28.7 56.6 11.8 35.0

To illustrate Table 10's results, consider the predicted probabilities of an


identical man and woman reporting overall job satisfaction of seven, or of six or
seven. The baseline characteristics are the same as those used to calculate the
predicted probabilities in Section 3, with the addition of the following information
on work values: the first most important aspect of a job is pay and the second most
important aspect is hours of work. From the estimates reported in Table 10, a
woman with these characteristics has a 35.8% estimated probability of reporting
overall job satisfaction of 7 and a 64.2% probability of reporting overall job
satisfaction of 6 or 7; for the identical man, these estimated probabilities are
markedly lower at 18.7 and 46.4%, respectively.
Table 11 illustrates how these predicted probabilities change with certain
individual and job characteristics. Higher education reduces women's estimated
job satisfaction but has little effect (relative to that of medium education) on
men's. Changing from married to never married also reduces women's estimated
job satisfaction, while reducing weekly hours of work raises it; both of these
changes have no effect on men's estimated job satisfaction. Last, removing
promotion opportunities reduces both men's and women's estimated job satisfac-
tion substantially; the woman now having an estimated less than one in three
probability of reporting job satisfaction of 7, with that of an identical man being
less than one in eight.

8. Conclusion

This paper has used a large-scale British data set to analyse the relationship
between gender and job satisfaction. The first finding is that women report
significantly higher levels of most kinds of job satisfaction than do men, even
when a large number of individual and job characteristics are controlled for.
Variables reflecting work values, which differ substantially between men and
women and which are important predictors of job satisfaction in their own right,
do not explain women workers' higher satisfaction, nor seemingly does sample
selection.
A.E. Clark~Labour Economics 4 (1997) 341-372 365

The paper's preferred explanation of women's higher satisfaction with what are
objectively worse jobs appeals to the idea, common in other social sciences, of
relative utility or well-being. Women's higher job satisfaction does not reflect that
their jobs are unobservedly better than men's, but rather that, perhaps because
their jobs have been so much worse in the past, they have lower expectations. To
test this, first an explicit variable measuring the income against which workers
might compare their own was introduced. Although this helps to explain reported
levels of job satisfaction, the gender differential remains significant. A second,
more general, approach was to analyse the relationship between the gender
satisfaction differential and variables which are likely correlated with workers'
expectations. The gender differential disappears for younger workers, higher-
educated workers, those in professional or managerial positions, those whose
mothers had a professional job, and those working at male-dominated workplaces.
These groups are all likely to have higher expectations about what their jobs
should entail.
The expectations interpretation suggests that women's higher job satisfaction
may be a transitory phenomenon, caused by women's improved position in the
labour force relative to their expectations. Once women's labour market rewards
stop improving (or, more accurately, only improve at the same rate as those of
men) men's and women's reported satisfaction should be identical, once other
individual and job attributes are controlled for, as there will be no gender
difference in rewards relative to expectations. No direct test of the relative utility
hypothesis has been carried out; nor is it likely that any definitive test can be
applied, as reliable information on expectations is rarely collected. However, with
the well-documented relationship between job satisfaction and labour market
behaviour, the understanding of what makes different groups of workers satisfied
should be given a high priority by labour economists.

Acknowledgements

I am grateful to Nick Buck, Shirley Dex, Peter Dolton, Steinar Holden, Edwin
Leuven, Ross McKay, Andrew Oswald, Sol Polachek, Arthur van Soest, Peter
Warr, Frances Woolley and seminar participants at Buckingham, CREST, the
Econometric Society European Meeting (Maastricht), the EMRU Labour Eco-
nomics Study Group, Essex, Guelph, Kent, Leiden, the Out of the Margin
Conference (Amsterdam), Tilburg and Warwick for advice and comments. The
data used in this paper were made available through the ESRC Data Archive. The
data were originally collected by the ESRC Research Centre on Micro-social
Change at the University of Essex. Neither the original collectors of the data nor
the Archive bear any responsibility for the analyses or interpretations presented
here. Part of this research was carried out at DELTA, Paris, whose hospitality is
gratefully acknowledged. This work was supported by the European Union. The
366 A.E. Clark~Labour Economics 4 (1997) 341-372

views expressed in this article are those of the author and do not represent those of
the OECD or of its member countries.

Appendix A. The Distribution of Job Satisfaction

See Table 12.

A p p e n d i x B. Variables and Definitions

Male dummy: Respondent is male.


Age: Age of respondent at date of interview.
Health dummies (3): Respondents classify their own health, compared to people
of their own age. Categories: excellent; good; fair to very poor. Omitted category:
fair to very poor.
Education dummies (3): 'High', Degree, teaching qualification or other higher
qualification; 'Medium', Nursing qualification, A-levels, O-levels or equivalent;
'Low', neither of the above. Omitted category: 'Low'.
Marital Status dummies (6): Married; Cohabiting; Widowed; Divorced; Sepa-
rated; Never Married. Omitted category: Never Married.
Renter dummy: Respondent lives in rented or rent-free accommodation.
Log income: Natural log of usual monthly gross pay from respondent's main
job.
Log hours: Natural log of usual monthly hours (excluding overtime).
Travel Time: Number of minutes taken to get to work each day.
Manager dummy: Respondent has some managerial or supervisory duties.
Union member dummy: Respondent is a member of a recognized union at their
workplace.
Temporary or contract work dummy: Respondent's current job is seasonal,
temporary, casual or a job done under contract or for a fixed period of time.

Table 12
1 (%) 2 (%) 3 (%) 4 (~) 5 (%) 6 (%) 7 (%) N
Overall 3.4 1.8 4.5 12.5 19.1 26.9 31.9 5142
Promotion 17.0 4.6 6.3 29.4 11.6 9.4 21.7 4541
Pay 12,9 4,8 10.2 20.2 17.1 14,1 20.8 5132
Relations 4.6 2,3 3.3 13.0 11.9 19.6 45.2 5061
Job security 9.0 3.4 5.4 15,6 12.0 16.3 38.4 5082
Initiative 3.5 1,6 2.9 10.2 12.2 21.0 48.6 5120
Work itself 3.5 1.3 3,6 13.2 14.7 22.1 41.7 5144
Hours 5.1 3.0 6.0 18,1 13.7 16.6 37.4 5141

Note: Unweighted data.


A.E. Clark/Labour Economics 4 (1997) 341-372 367

Table 13
Male 0.471 (0.134)
Age 0.084 (0.011)
Age-squared/1000 - 1.182 (0.135)
Black - 0 . 1 2 0 (0.142)
Asian subcontinent - 0 . 5 9 3 (0.124)
Region dummies
Health: excellent 0.238 (0.048)
Health: good 0.254 (0.044)
Education: higher 0.397 (0.050)
Education: A / O / n u r s i n g 0.285 (0.042)
Female, cohabiting 0.211 (0.103)
Female, widowed 1.138 (0.176)
Female, divorced 0.840 (0.127)
Female, separated 0.751 (0.173)
Female, never married 0.566 (0.l 14)
Male, cohabiting - 0.342 (0.098)
Male, widowed 0.126 (0.289)
Male, divorced - 0.230 (0.142)
Male, separated - 0.503 (0.239)
Male, never roamed - 0.206 (0.098)
Female, spouse's pay < £500 pcm - 0.407 (0.208)
Female, spouse's pay £500-£1000 pcm 0.089 (0.169)
Female, spouse's pay £1000-£1500 pcm - 0 . 1 5 4 (0,170)
Female, spouse's pay > £1500 pcm - 0 . 5 2 7 (0.176)
Male, spouse's pay < £500 pcm 0.117 (0.358)
Male, spouse's pay £500-£1000 pcm 0.031 (0.354)
Male, spouse's pay £1000-£1500 pcm - 0.209 (0.363)
Male, spouse's pay > £1500 pcm - 0.393 (0.379)
Female, one child in household - 0 . 2 1 7 (0.086)
Female, two children in household - 0.273 (0.105)
Female, three children in household - 0 . 4 2 9 (0.145)
Female, four + children in household - 0.637 (0.255)
Male, one child in household - 0.162 (0.097)
Male, two children in household - 0 . 2 0 4 (0.116)
Male, three children in household - 0 . 4 8 3 (0.158)
Male, four + children in household - 0.370 (0.259)
Children present aged 0 - 2 - 0.487 (0.076)
Children present aged 3 - 4 - 0.257 (0.072)
Children present aged 5-11 - 0.038 (0.068)
Children present aged 12-15 o.145 (0.060)
Children present aged 16-18 - 0.216 (0.065)
Annual rest of household income/10000 0.162 (0.023)
Female, shopping: shared/self 0.059 (0.084)
-

Female, cooking: shared/self - O. 162 (0.097)


Female, cleaning: shared/self - 0.462 (0.096)
Female, washing: shared/self 0,380 (0.109)
Male, shopping: shared/self 0.135 (0.060)
Male, cooking: shared/self - 0.209 (0.065)
368 A.E. Clark/Labour Economics 4 (1997) 341-372

Table 13 (continued)
Male, cleaning: shared/self - 0,142 (0.064)
Male, washing: shared/self 0.064 (0,075)
Renter -0.185 (0.058)
House buyer 0.155 (0.053)
No. of others employed in household: 1 0.774 (0,146)
No. of others employed in household: 2 0.458 (0.139)
No. of others employed in household: 3 - 0.287 (0.138)
No. of others employed in household: 4 + -0,133 (0.148)
Annual unearned income/1000 - 0 . 2 8 7 (0,013)
Female, care giver in household - 0 . 1 9 0 (0.129)
Female, care giver outside of household 0.048 (0.070)
Male, care giver in household - 0 . 0 4 6 (0.140)
Male, care giver outside of household 0.I29 (0.090)
Female, professional parent - 0.163 (0.055)
Male, professional parent -0.135 (0.055)
Female, spouse's hours less than 16 p,w. 0.513 (0.302)
Female, spouse's hours between 16 and 29 p.w. 0.665 (0.292)
Female, spouse's hours between 30 and 39 p,w. 0.996 (0.187)
Female, spouse's hours 40 p.w. or more 0.783 (0.178)
Male, spouse's hours less than 16 p.w. 0,240 (0.372)
Male, spouse's hours between 16 and 29 p.w. 0.332 (0.365)
Male, spouse's hours between 30 and 39 p,w. 0357 (0,357)
Male, spouse's hours 40 p.w. or more 0,096 (0.358)
Constant - 1.801 (0.283)
No. of observations used 7064

Standard errors in parentheses.

Promotion opportunities dummy: Respondent has opportunities for promotion


in their current job.
Job tenure: Number of months spent in current job.
Second job dummy: Respondent earns money from a second job.
Establishment size dummies (3): Number of workers at establishment is < 25;
25-199; 200 + . Omitted category: 200 + .
Region dummies (18): Standard British regions plus seven metropolitan areas.
Industry dummies (10): Agriculture, forestry and fishing; energy and water
supplies; extraction of minerals and ores other than fuels, manufacture of metals,
mineral products and chemicals; metal goods, engineering and vehicles industries;
other manufacturing industries; construction; distribution, hotels and catering;
transport and communication; banking, finance, insurance, business services and
leasing; other services. Omitted category: Other services.
Occupation dummies (9): Managers and administrators; professional; associate
professional and technical; clerical and secretarial; craft and related; personal and
protective service; sales; plant and machine operative; other. Omitted category:
Other.
A.E, Clark/Labour Economics 4 (1997) 341-372 369

Table 14
Age 20-29 0.44 (14.2)
Age 30-39 0.52 (15.4)
Age 40-49 0.53 (14.7)
Age 50-59 O.52 (13.3)
Age 60 + 0.37 (7.7)
Male 0.17 (8.7)
Health dummies (5) l,a
Regional dummies (18) I,d
Education dummies (13)
Accident dummies (3) vd
Marriage dummies (6)
Industry dummies (61) l,a
Occupation dummies (76)
When work dummies (10) 1,t
Establishment size dummies (10)
Sex mix at work dummies (5)
Organization type dummies (7)
Temporary contract - 0.23 (8.9)
Pay includes incentives 0.08 (4.9)
Part-time - 0.69 (31.4)
Union member 0.06 (3,2)
Trade Union recognised 0.04 (2.1)
Manager 0.15 (9.6)
Pension member 0.12 (7.1)
Job tenure - 5.4E-07 (0,1)
Job tenure squared - 1.8E-12 (0.1)
Constant 6.43 (76,6)
Number of observations 4099
Adjusted R 2 78%

T-statistics in parentheses.

Appendix C. Employment Probit

See Table 13.

Appendix D. Gross Monthly Income Equation

See Table 14.

References

Adams, J.S., 1963. Toward an understanding of inequity. Journal of Social Psychology 67, 422-436.
Akerlof, G.A., Rose, A.K., Yellen, J.L, 1988. Job switching and job satisfaction in the U.S labor
market. Brookings Papers on Economic Activity 2, 495-582.
370 A.E. Clark/Labour Economics 4 (1997) 341-372

Arvey, R.D., Abraham, L.M., Bouchard, T.J., Segal, N.L., 1989. Job satisfaction: Environmental and
genetic components. Journal of Applied Psychology 74, 187-192.
Argyle, M., 1989. The Psychology of Happiness, Routledge, London.
Bell, D.E., 1982. Regret in decision making under uncertainty. Operations Research 30, 961-981.
Bell, D.E., 1985. Disappointment in decision making under uncertainty. Operations Research 33, 1-27.
Birke, L., 1986. Women, Feminism and Biology, Wheatsheaf Books, Brighton.
Blanchflower, D.G., Oswald, A.L, 1992. Entrepreneurship and Supernormal Returns: Evidence from
Britain and the US, Mimeo, Dartmouth College.
Blanchflower, D.G., Oswald, A.J. Warr, P.B., 1993. Well-Being Over Time in Britain and the USA,
Mimeo, Dartmouth College.
Blau, F.D., Kahn, L.M., 1992. The gender earnings gap: Learning from international comparisons.
American Economic Review 82, 533-538.
Brown, C., Medoff, J., 1989. The employer size-wage effect. Journal of Political Economy 97,
1027-1059.
Brown Johnson, N., Bobko, P., Hartenian, L.S., 1992. Union influence on local union leaders'
perceptions of job insecurity: An empirical test. British Journal of Industrial Relations 30, 45-60.
Buck, N., Gershuny, J., Rose, D., Scott, J. (Eds.), 1994. Changing Households, ESRC Research Centre
Press, Colchester.
Clark, A.E., 1996a, Are Wages Habit-Forming? Evidence from Micro Data, Mimeo, OECD.
Clark, A.E., 1996b. L'utilit6 est-elle relative? Analyse h l'aide de donntes sur les mtnages. (Is utility
relative? Evidence from household data.). Economie et Prtvision 121, 151-164.
Clark, A.E., 1996c. Job satisfaction in Britain. British Journal of Industrial Relations 34, 189-217.
Clark, A.E., Oswald, A.J., 1994. Unhappiness and unemployment. Economic Journal 104, 648-659.
Clark, A.E., Oswald, A.J., 1996. Satisfaction and comparison income. Journal of Public Economics 61,
359-381.
Clark, A.E., Oswald, A.J., Warr, P.B., 1996. Is job satisfaction U-shaped in age?. Journal of
Occupational and Organizational Psychology 69, 57-81.
Crawford, M., 1995. Talking Difference. On Gender and Language, Sage Publications, London.
Crosby, F.J., 1982. Relative Deprivation and Working Women, Oxford University Press, Oxford.
Clegg, C.W., 1983. Psychology of employee lateness, absence and turnover: A methodological critique
and an empirical study. Journal of Applied Psychology 68, 88-101.
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, Dartmouth, Aldershot.
Duesenberry, J.S., 1949. Income, Saving and the Theory of Consumer Behaviour, Harvard University
Press, Cambridge, MA.
Elster, J., Roemer, J., 1991. Introduction. In: Elster, J., Roemer, J. (Eds.), Interpersonal Comparisons of
Well-Being, Cambridge University Press, Cambridge.
Equal Opportunities Commission, 1992. Pay and Gender in Britain: 2, IRS, London.
Freeman, R.B., 1978. Job satisfaction as an economic variable. American Economic Review 68,
135-141.
Garai, J.E., Scheinfeld, A., 1968. Sex differences in mental and behavioral traits. Genetic Psychology
Monographs 77, t69-299.
Goldberg, D.P., 1972. The Detection of Psychiatric Illness by Questionnaire, Oxford University Press,
Oxford.
Greene, W., 1993. Econometric Analysis, 2nd ed., Macmillan, New York.
Groot, W., H. Maassen van den Brink, 1996. Job Satisfaction, Wages and Allocation of Older Workers,
Tinbergen Institute, Discussion Paper TI 96-73/3.
Gruenberg, B., 1980. The happy worker: An analysis of educational and occupational differences in
determinants of job satisfaction. American Journal of Sociology 86, 247-271.
Hakim, C., 1991. Grateful slaves and self-made women: Fact and fantasy in womens work orientations.
European Sociological Review 7, 101-121.
A.E. Clark/Labour Economics 4 (1997) 341-372 371

Hagenaars, A.J., 1986. The Perception of Poverty, North-Holland, Amsterdam.


Hamermesh, D.S., 1977. Economic aspects of job satisfaction. In: Ashenfelter, O.C, Oates, W.E.
(Eds.), Essays in Labor Market Analysis, John Wiley, New York.
Harsanyi, J.C., 1986. Interpersonal utility comparisons. In: Eatwell, J., Milgate, M., Newman, P.
(Eds.), The New Palgrave Dictionary of Economics, Macmillan, London.
Hodson, R., 1985. Workers' comparisons and job satisfaction. Social Science Quarterly 66, 266-280.
Homans, G.C., 1961. Social Behaviour: Its Elementary Forms, Harcourt Brace, New York.
Idson, T.L., 1990. Establishment size, job satisfaction and the structure of work. Applied Economics
22, 1007-1018.
Janson, P., Martin, J.K., 1982. Job satisfaction and age: A test of two views. Social Forces 60,
1088-1102.
Kahneman, D., Tversky, A., 1979. Prospect theory: An analysis of decision under risk. Econometrica
47, 263-291.
Kapteyn, A., Van Herwaarden, F.G., 1980. Interdependent welfare functions and optimal income
distribution. Journal of Public Economics 14, 375-397.
L~vy-Garboua, L., Montmarquette, C., 1994. On reported job satisfaction: A test of subjective
well-being models and a new interpretation, Universit6, Paris I, mimeo.
Lewis, J., 1992. Women in Britain Since 1945, Blackwell, Oxford.
Loomes, G., Sugden, R., 1982. Regret theory: An alternative theory of rational choice under
uncertainty. Economic Journal 92, 805-824.
Loprest, P.J., 1992. Gender differences in wage growth and job mobility. American Economic Review
82, 526-532.
Loscocco, K.A., Spitze, G., 1991. The organizational context of women's and men's pay satisfaction.
Social Science Quarterly 72, 3-19.
Major, B., 1987. Gender, justice and the psychology of entitlement. In: Shaver, P., Hendrick, C. (Eds.),
Sex and Gender, Sage Publications, Newbury Park.
Mangione, T.W., Quinn, R.P., 1975. Job satisfaction, counterproductive behavior and drug use at work.
Journal of Applied Psychology 60, 114-116.
McEvoy, G.M., Cascio, W.F., 1985. Strategies for reducing employee turnover: A meta-analysis.
Journal of Applied Psychology 70, 342-353.
Meng, R., 1990. The relationship between unions and job satisfaction. Applied Economics 22,
1635-1648.
Miller, J., 1980. Individual and occupational determinants of job satisfaction. Sociology of Work and
Occupations 7, 337-366.
Moir, A., Jessel, D., 1989. Brainsex, Michael Joseph, London.
Morris, J.M., Villemez, W.J., 1992. Mobility potential and job satisfaction. Work and Occupations 19,
35-58.
Neil, C.C., Snizek, W.E., 1987. Work values, job characteristics and gender. Sociological Perspectives
30, 245-265.
Van Praag, B.M., 1991. Ordinal and cardinal utility. Journal of Econometrics 50, 69-89.
Riach, P.A., Rich, J., 1987. Testing for sexual discrimination in the labour market. Australian
Economic Papers 26, 165-178.
Rose, S., Lewontin, R.C., Kamin, L.J., 1984. Not in Our Genes, Penguin Books, Harmondsworth.
Ross, C.E., Reskin, B.F., 1992. Education, control at work and job satisfaction. Social Science
Research 21, 134-148.
Runciman, W.G., 1966. Relative Deprivation and Social Justice, Routledge and Kegan Paul, Henley.
Schmidt, C.M., Zimmerman, K.F., 1991. Work characteristics, firm size and wages. Review of
Economics and Statistics 73, 705-710.
Sloane, P., Battu, HL, Seaman, P.T, 1995. Overeducation, Undereducation and the British Labour
Market, Mimeo, University of Aberdeen.
372 A.E. Clark/Labour Economics 4 (1997) 341-372

Sloane, PJ., Williams, H., 1995. Job Satisfaction, Comparison Income and Gender, Mimeo, University
of Aberdeen.
Stouffer, S.A. et al., 1949, The American soldier, Princeton University Press: Princeton.
Townshend-Smith, R., 1989. Sex Discrimination in Employment: Law, Practice and Policy, Sweet and
Maxwell, London.
de Vaus, D., McAllister, I., 1991. Gender and work orientation. Work and Occupations 18, 72-93.
Waldman, M., 1994. Systematic errors and the theory of natural selection. American Economic Review
84, 482-497.
Watson, R.I., Lindgren, H.C., 1973. Psychology of the Child, 3rd ed., John Wiley, New York.
Woittiez, I., Theeuwes, J., 1994. Welfare Distributions of Dutch Elderly Based on Happiness Rankings,
Mimeo, Leiden University.
Wright, R.E., Ermisch, J.F., 1991. Gender discrimination in the British labour market: A reassessment.
Economic Journal 101,508-522.
Zavoina, R,, McKeivey, W., 1975, A Statistical Model for the Analysis of Ordinal-Level Dependent
Variables, Journal of Mathematical Sociology, Summer, 103-120.

You might also like