The two major variables used to operationalize socioeconomic

position in studies of social inequalities in health are social

stratification and social class. Social stratification refers to the
ranking of individuals along a continuum of economic or cultural
attributes such as income or years of education. These rankings
are known as simple gradational measures.
1
Most social epi-
demiologists use several measures of social stratification
simultaneously because single measures have been insufficient
to account for social inequalities in the health of populations.
Measures of social stratification are important predictors of pat-
terns of mortality and morbidity,
2
and during the last decade,
a number of investigations of social inequalities in health have
assessed the relation between indicators of social stratification and
health outcomes. However, despite their usefulness in predicting
health outcomes, these measures do not reveal the social mech-
anisms that explain how individuals arrive at different levels
of economic, political, and cultural resources,
3
in part perhaps
IJE vol.32 no.6 International Epidemiological Association 2003; all rights reserved. International Journal of Epidemiology 2003;32:950958
DOI: 10.1093/ije/dyg170
SPECIAL THEME: MENTAL HEALTH
The associations of social class and social
stratification with patterns of general and
mental health in a Spanish population
Carles Muntaner,
1
Carme Borrell,
2
Joan Benach,
3
M Isabel Pasarn
2
and Esteve Fernandez
4
Accepted 28 February 2003
Background Social class, as a theoretical framework, represents a complementary approach to
social stratification by introducing social relations of ownership and control over
productive assets to the analysis of inequalities in economic, political, and
cultural resources. In this study we examined whether measures of social class
were able to explain and predict self-reported general and mental health over
and above measures of social stratification.
Methods We tested this using the Barcelona Health Interview Survey, a cross-sectional
survey of 10 000 residents of the citys non-institutionalized population in 2000.
We used Erik Olin Wrights indicators of social class position, based on ownership
and control over productive assets. As measures of social stratification we used
the Spanish version of the British Registrar General (BRG) classification, and
education. Health-related variables included self-perceived health and mental
health as measured by Goldbergs questionnaire.
Results Among men, high level managers and supervisors reported better health than all
other classes, including small business owners. Low-level supervisors reported
worse mental health than high-level managers and non-managerial workers,
giving support to Wrights contradictory class location hypothesis with regard to
mental health. Social class indicators were less useful correlates of health and
mental health among women.
Conclusions Our findings highlight the potential health consequences of social class positions
defined by power relations within the labour process. They also confirm that
social class taps into parts of the social variation in health that are not captured
by conventional measures of social stratification and education.
Keywords Social class, social stratification, socioeconomic status, mental health, self-rated health
1
Department of Behavioral and Community Health Nursing and Department
of Epidemiology and Preventive Medicine, University of Maryland at
Baltimore, USA.
2
Municipal Institute of Public Health of Barcelona, Barcelona, Spain.
3
Universitat Pompeu Fabra, Barcelona, Spain.
4
Institut Catal dOncologia, Barcelona, Spain.
Correspondence: Carles Muntaner Bonet, Suite 645/BCH, University of
Maryland-Baltimore, 655 West Lombard Street, Baltimore, MD 21201, USA.
E-mail: cmunt001@umaryland.edu
950

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

because they have generally been selected for pragmatic
considerations, i.e. availability of data, rather than for theoretical
reasons.
Social class is defined by relations of ownership or control over
productive resources (i.e. physical, financial, organizational).
Social class has important consequences for the lives of indi-
viduals: the extent of an individuals legal right and power to
control productive assets determines an individuals strategies
and practices devoted to acquire income and, as a result, deter-
mines the individuals standard of living.
1
Thus the class position
of business owner compels its members to hire workers and
extract labour from them, while the worker class position
compels its members to find employment and perform labour.
Although there have been few empirical studies of social class
and health, the need to study social class has been noted by
social epidemiologists.
2,4
Social class provides an explicit
relational mechanism (property, management) that explains
how economic inequalities are generated and how they may
affect health. For example, in a recent study,
5
a team of US
epidemiologists found that low-level supervisors, who could
hire and fire front line personnel but did not have policy or
decision-making authority in the firm, showed higher rates of
depression and anxiety disorders than both upper management
(who had authority and decision-making attributes) and non-
management workers (who had neither). This finding was
predicted by the contradictory class location hypothesis (super-
visors are in conflict with both workers and upper management
and do not have control over policy) but was not predicted or
explained by indicators of years of education or income gradients.
Moreover, the income hypothesis would have failed to provide
a mechanism and would have led to the expectation that super-
visors, because of their higher incomes, would present lower
rates of anxiety and depression than workers. A handful of
studies in psychiatric epidemiology
57
suggest that social strati-
fication and social class are not equivalent; rather, they capture
different parts of the social variation in population mental
health. Therefore, the purpose of our study was to examine the
relationships between measures of social stratification (education,
British Registrar General Classification [BRG]), measures of
social class (Wrights social class indicators, i.e. relationship to
productive assets), and indicators of general health and mental
health.
The measures of social class used in our investigation
originate from a social class model that has been accumulating
empirical support over the last 20 years (e.g. refs 814). Wrights
social class indicators assess ownership of productive assets, and
control and authority relations in the workplace (control over
organizational assets
1
). Property rights over the financial or
physical assets used in the production of goods and services
generate three class positions: employers, who are self-
employed and hire labour; the traditional petit bourgeoisie, who
are self-employed but do not hire labour; and workers who sell
their labour.
1
These social class positions reflect the relational
properties underlying economic inequality.
15
Indicators of
productive asset ownership gauge a relational mechanism that
generates economic inequality (i.e. deriving income from
owning property). Both neo-material
16
and psychological
5
mechanisms suggest that owners might present better overall
health and mental health than workers. Large property owners
tend to be wealthier
17
than others and thus might be expected
to experience the greater material well-being that is conducive
to better health.
3
In addition, large owners enjoy the predict-
ability and control in life that are predictive of better mental
health.
18
They are not subject to the stressors of unpredict-
ability and lack of control associated with relying exclusively on
salaries or wages for income. As a result, they may enjoy better
health. Even small property owners can derive economic security
from wealth, which is more concentrated among property
owners than income.
19,20
However, since most small businesses
go bankrupt, the suitability of this hypothetical mechanism to
small capitalist class positions is less evident.
21
These hypothetical
mechanisms linking property ownership to economic security
were part of the underlying rationale for this study.
Control over organizational assets (power and control in the
workplace) is determined by two kinds of relations at work:
(1) influence over company policy (e.g. making decisions over
number of people employed, products or services delivered,
amount of work performed, size and distribution of budgets);
and (2) sanctioning authority (granting or preventing pay raises
or promotions, hiring, firing, or temporally suspending a sub-
ordinate).
1
The supervisory and policy making functions of
managers allow them to enjoy greater wealth than workers,
for example, through income derived from shares of stock,
incentives, bonuses, and hierarchical pay scales.
22
As a conse-
quence, we anticipate that managers will present better health
and mental health than non-managers, in accordance with the
hypothesis derived from asset ownership. Furthermore,
workplace authority relations add another mechanism that may
impact health, i.e. control over ones work and the ability to
extract labour effort from others, increasing ones sense of control
and predictability at work.
23
Indeed, the work organization
literature,
23
including the Whitehall study,
24
suggests that in
addition to greater access to income, wealth, and job security,
control over work may be a mechanism linking managerial class
positions to better health.
Following Erik Wrights class theory, we defined and
measured managerial class positions according to policy-making
power within the labour process and supervisory functions over
others labour. Social stratification (e.g. occupational categories)
does not define or measure relational mechanisms within the
labour process. Popular stratification measures such as occu-
pational groups cannot generate specific hypotheses because
they are compatible with many potential mechanisms (e.g.
occupational prestige categories, income, authority).
According to Wrights contradictory class locations hypoth-
esis, supervisors are in a special position (i.e. a contradictory
class position in production relations
9
), subjected both to the
pressure of upper management to discipline the workforce and
the antagonism of subordinate workers, while exerting little
influence over company policy.
1
This situation may expose
supervisors to high demands and low control at work, which
are risk factors for mental disorders.
5
Therefore, supervisors are
more likely to present poorer mental health than managers.
Wright includes skills/credentials relations as part of his map
of class positions (Figure 1, the expert, semi-skilled, and un-
skilled class positions
1
). Experts are defined as those holding
jobs that require skills, particularly accredited credentialed
skills, which are scarce relative to their demand by the market.
Experts enjoy a credential rent: their wages are usually above
the cost of the reproduction of their training.
9
Semi-skilled and
SOCIAL STATUS AND HEALTH IN SPAIN 951

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

952 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
unskilled class positions are defined as jobs requiring skills that
are in large supply, particularly un-credentialed skills. Because
credentials provide access to labour markets with higher pay and
less hazardous working conditions, experts would be expected to
have better health status than semi-skilled and unskilled
workers.
Our interest lay in the public health significance of two social
class relations in populationsownership and managerial
control. We hypothesized that both positions would confer
unique protection against ill health. In addition, we predicted
that supervisors would exhibit worse health than either
managers or workers. Finally, we hypothesized that experts
would have better health than less-skilled workers. Both self-
reported health, a valid predictor of mortality,
25
and general
mental health
26
were optimal for our purposes as they capture
a populations average physical and mental health
27
rather than
the prevalence of specific conditions or high risk. Furthermore,
both indicators have been associated with social risk factors
and are useful for needs assessment and public health
interventions.
28
Methods
Study population, sample, and data collection
The sampling frame was the 2000 non-institutionalized popu-
lation of Barcelona city (1 500 000 inhabitants, Catalonia,
Spain). Data were collected as part of the 2000 Barcelona Health
Interview Survey (HIS), a cross-sectional population survey
carried out approximately every 5 years since 1983.
29
We
generated a representative stratified sample of the 10 000 non-
institutionalized residents using the 10 districts of the city as
strata. In each stratum a random sample of residents was
obtained. With a sample of 10 000, the alpha error was 4.5%
and the maximum global error 1% (this global error is one-half
the width of the desired sample CI). The information was
collected through face-to-face interviews carried out at home
between February 2000 and March 2001. The study reported
here included the working population 1664 years of age
(n = 4219). The age distribution of our sample closely matches
the age distribution obtained in the 2001 survey of the
Barcelona employed population.
30
Health and mental health variables
Perceived health status was measured through a single
question: Would you say your overall health is very good,
good, fair, bad or very bad? A dichotomous outcome measure
was created with fair, bad or very bad coded 1 and good and
very good coded zero (i.e. 1 = fair, bad, or very bad; 0 = very
good, good). Mental health was measured with the 12-item
version of the General Health Questionnaire (12-GHQ
26
).
Responses were summed and those scoring 3 were classified
as having a high probability of a psychiatric disorder.
We dichotomized the two variables for the following reasons:
(1) previous studies have dichotomized self-rated health
25
as
well as Goldbergs GHQ (we used the standard cut-off score
26
),
which renders our approach consistent with the literature;
(2) research on the cognitive aspects of survey responses shows
that respondents tend to dichotomize evaluations of their own
health status;
27
(3) dichotomized self-reports of health have
good predictive validity;
25
and (4) low cell frequency in the
intermediate categories of self-rated health advised towards
dichotomizing for the analysis.
Social class and social stratification variables
Erik Wrights map of class positions is presented in Figure 1.
Indicators for each of the 12 class positions were obtained from
a set of survey questions. Class positions in the property dimen-
sion were obtained through two questions inquiring whether
the respondent was self-employed and, if self-employed, the
number of people working for her. Self-employed who hired at
most one single worker were considered to occupy the petit-
bourgeois class position; self-employed having between 2 and
10 workers were defined as small employers; and self-employed
with more than 10 workers occupied the capitalist class
position.
1
Class positions in the organizational control dimension
were determined by questions assessing whether the respondent
worked as a manager (i.e. gerentes and directores) and what
the authority relations were in the workplace (i.e. the number
of workers supervised), yielding three class categories: managers
(those with the power to influence company policy and super-
vise one or more subordinates); non-managerial supervisors
(those with the power to supervise one or more subordinates
only); and non-managerial workers (those with no power
Figure 1 Erik O Wright class locations

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

SOCIAL STATUS AND HEALTH IN SPAIN 953
as defined above).
1
Information on the skill dimension (i.e.
experts, semi-skilled workers, and unskilled workers) was
obtained through the occupation and educational credentials of
those interviewed.
1
Professionals, university professors, man-
agers with a university degree, and technicians with a university
degree were considered experts. Managers, technicians, non-
university teachers, craftsmen, tradesmen with university
degrees, and clerks with a university degree were considered
semi-skilled. Other occupations were considered unskilled.
This measure of social class has been previously used in Spanish
surveys.
31
In multivariate analyses, managers and expert super-
visors were combined due to the small number of respondents
in managerial class positions.
As a measure of social stratification, we used the Spanish
adaptation of the British Registrar General Classification (BRG;
which includes five strata from BRG I to BRG V). The Spanish
version of the BRG was developed by comparing occupations in
Britain and Spain.
32
In most instances occupations fell into the
same stratum. However, in a few cases British occupations were
assigned to different social strata in Spain. For example, writers
and journalists in Spain were located in BRG I. Non-manual
occupations were assigned to BRG III and manual occupations
to BRG IV (whereas in the British classification, both manual
and non-manual skilled occupations are part of BRG III). The
Spanish version of the BRG has been widely used in Spain and
is the measure of social stratification recommended by the
Spanish Epidemiological Society.
29,32
In this study the respond-
ents occupation was used to assign each interviewee in the
sample to one of the five strata.
We used the highest completed level of education as another
measure of social stratification. Education was grouped in the
following strata: illiterate or no education, which included
people with 04 years of schooling, primary education (511
years of schooling), secondary education (1215 years of
education), and university or graduate school (16 years of
schooling).
Data analysis
Age-standardized percentages by social class and educational
levels were calculated using the direct method for each health-
related variable, with the whole study population as reference.
Logistic regression models were adjusted by age (continuous)
to calculate the association with health-related variables.
Reference categories for the odds ratios (OR) were class I for the
BRG classification, university or graduate studies for education,
and managers and expert supervisors for the Wright class
positions. We chose these positions as reference categories
because they had lower percentages of poor health and poor
mental health than other positions.
First, we obtained separate regression models for each of the
social class and social stratification variables. Then we generated
two multivariable-adjusted models, one with age, Wrights
social class indicators, and educational level, and the other
with age, Wrights class indicators, and the BRG classification.
Educational level and BRG classification were not included
simultaneously in the models because of their high cor-
relation. Goodness of fit was obtained using the Hosmer
and Lemeshow test.
33
All analyses included weights derived
from the complex sample design and were stratified by
gender.
Results
The study sample included 2345 men and 1874 women. As
Table 1 shows, 10.7% of men and 14.2% of women reported
poor health status; almost 9% of men and 15% of women
reported poor mental health. The majority of participants were
45 years, had achieved secondary education or a university
degree and belonged to BRG occupational classes III and IV. The
distributions of the BRG social stratifications are not comparable
to those in the general population because only the employed
(1664 years of age) were included in this study. Because older
Spaniards fall into lower strata than younger cohorts,
31,32
this
resulted in an over-representation of BRG I participants and an
under-representation of BRG II participants in our sample.
Using Wrights social class positions, semi-skilled workers
represented one-fifth of the population and unskilled workers
represented 31% of men and 47% of women.
Tables 2 and 3 show the health-related variables by social
class and social stratification in men and women. Men who had
less education and belonged to BRG classes IV and V had higher
percentages of poor perceived health and poor mental health.
Table 1 Description of the population studied by sex. Number of cases
and column percentages. Working population, 1664 years old.
Barcelona 2000
Males Females
(n = 2345) (n = 1873)
Variables n % n %
Age group
1624 246 10.5 235 12.5
2534 612 26.1 526 28.1
3544 615 26.2 559 29.8
4554 540 23.0 347 18.5
5564 332 14.2 206 11.0
Educational level
University 750 32.0 646 34.5
Secondary 731 31.2 576 30.8
Primary 742 31.6 557 29.7
Less than primary 119 5.1 94 5.0
Missing 4 0.2
British Registrar General classification
I 437 18.6 264 14.1
II 298 12.7 233 12.4
III 716 30.5 674 36.0
IV 768 32.7 486 25.9
V 120 5.1 214 11.4
Missing 7 0.3 3 0.1
EO Wright classification
Capitalists 46 2.0 16 0.9
Small employers 186 7.9 101 5.4
Petit bourgeoisie 195 8.3 112 6.0
Managers experts 60 2.5 23 1.2
Managers semi-skilled 32 1.4 6 0.3
Managers unskilled 0 0
Supervisors experts 113 4.8 60 3.2
Supervisors semi-skilled 142 6.1 51 2.7
Supervisors unskilled 120 5.1 78 4.1
Workers experts 204 8.7 159 8.5
Workers semi-skilled 505 21.5 379 20.2
Workers unskilled 735 31.3 878 46.8
Missing 7 0.3 11 0.6
Health-related variables
Poor perceived health status 249 10.7 265 14.2
Poor mental health 208 8.9 281 15.0

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Using Wrights social class indicators, those in capitalist class
positions did not have better health. The best health indicators
were found for expert and semi-skilled managers and for expert
supervisors. Poor perceived health varied from 2.7% among
expert supervisors to 14 % among unskilled workers; poor
mental health varied from 3.1% among semi-skilled managers
to 14.3% among unskilled supervisors. These percentages were
similar to age-standardized percentages (Table 2). Higher per-
centages of women with less education and those located in
BRG classes IV and V had poor perceived health and poor
mental health. Wrights social class positions showed a better
health profile for both managers and experts. Poor perceived
health was more frequent among unskilled workers (19%).
Mental health was worse among unskilled supervisors
(17.9%) (Table 3).
Tables 4 and 5 present the results from bivariate and multi-
variate logistic regression models for self-reported health and
mental health, respectively. Among men, those in BRG classes
III, IV, and V were more likely to have poor health than those
in the reference category of BRG class I; those with less than a
university degree were more likely to have poor health than
those with a university degree; and most class positions (except
for capitalists and expert workers) had worse health than the
reference category of expert and semi-skilled managers and
expert supervisors. In the multivariate models that included
social stratification (i.e. education or BRG) and Wrights social
class positions (Models 1 and 2 in Table 4), the majority of these
associations were retained; the OR was 4 for small employers,
semi-skilled and unskilled supervisors, and unskilled workers.
The addition of the stratification indicators to the models, includ-
ing social class, improved the model fit according to the
Likelihood Ratio Test (Table 4). Among women, using Wrights
class indicators, significant associations emerged for BRG classes
IIIV, education, and unskilled workers. In Models 1 and 2,
significant associations were present only for those with less
than primary education and those in BRG classes IV and V. The
addition of social stratification indicators improved the fit of the
social class model (Table 4).
Among men, secondary (i.e. high school) education was
associated with poor mental health. Unskilled supervisors
were almost three times (OR = 2.9, 95% CI: 1.36.6) as likely to
manifest poor mental health as managers and expert super-
visors, the reference category. In multivariate models the only
association that was significant was with unskilled supervisors.
The addition of social stratification indicators did not improve
the fit of the social class models. Among women, BRG classes IV
and V, low education, and most of Wrights social class positions
were significantly associated with poor mental health in
954 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 2 Distribution (number of participants, percentage, and age-
adjusted percentage) of health-related variables by measures of social
stratification and social class. Working men population, 1664 years
old. Barcelona 2000
Poor perceived Poor mental
health health
Age- Age-
Social stratification standard standard
and social class n % % n % %
Educational level
University 41 5.5 5.9 55 7.4 7.3
Secondary 64 8.8 9.4 75 10.3 10.4
Primary 117 15.9 15.2 68 9.2 9.7
Less than primary 28 23.5 15.8 10 8.4 11.3
British Registrar General classification
I 24 5.5 6.0 29 6.7 6.0
II 10 3.4 4.4 31 10.5 10.8
III 79 11.1 10.3 65 9.1 8.9
IV 114 15.0 14.5 69 9.0 9.3
V 20 16.8 16.9 14 11.6 13.3
EO Wright classification
Capitalists 2 4.3 4.4 4 8.7 7.4
Small employers 25 13.4 12.9 18 9.7 10.1
Petit bourgeoisie 24 12.4 8.1 18 9.2 8.1
Managers experts 2 3.4 2.6 2 3.4 3.0
Managers
semi-skilled 0 0 1 3.1 3.9
Managers
unskilled
Supervisors experts 3 2.7 4.8 8 7.1 6.6
Supervisors
semi-skilled 19 13.4 11.8 13 9.2 13.3
Supervisors
unskilled 15 12.8 14.0 17 14.3 15.1
Workers experts 9 4.4 5.2 14 7.0 7.1
Workers
semi-skilled 47 9.3 9.9 48 9.6 10.1
Workers
unskilled 102 14.0 14.3 67 9.2 9.2
Table 3 Distribution (number of subjects, percentage and age-
standardised percentage) of health-related variables by measures of
social stratification and social class. Working women population,
1664 years old. Barcelona 2000
Poor perceived Poor mental
health health
Age- Age-
Social stratification standard standard
and social class n % % n % %
Educational level
University 44 6.8 7.5 77 11.9 12.2
Secondary 70 12.2 13.7 101 17.6 17.3
Primary 109 19.6 17.8 81 14.6 14.9
Less than primary 42 44.7 31.4 21 22.3 19.4
British Registrar General classification
I 14 5.3 5.7 27 10.2 8.4
II 18 7.7 8.3 34 14.6 14.7
III 76 11.3 12.3 87 12.9 13.0
IV 89 18.4 20.0 87 18.0 17.4
V 69 32.4 26.3 45 21.0 20.8
EO Wright classification
Capitalists 2 12.5 14.0 2 11.8 16.0
Small employers 15 15.0 14.2 17 16.8 15.0
Petit bourgeoisie 15 13.5 21.5 15 13.4 11.7
Managers experts 0 0 0 1 4.5 5.0
Managers
semi-skilled 0 0 0 2 33.3 24.4
Managers
unskilled
Supervisors experts 6 10.0 9.9 3 5.0 8.2
Supervisors
semi-skilled 4 7.8 6.8 7 13.7 12.0
Supervisors
unskilled 12 15.4 13.9 14 17.9 21.6
Workers experts 6 3.8 4.6 16 10.1 8.7
Workers
semi-skilled 38 11.7 63 16.6 16.4
Workers
unskilled 166 19.0 19.5 141 16.1 16.0

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

bivariate regressions. However, in multivariate Model 1
(adjusted by education), the only associations that remained
significant were those involving semi-skilled workers and those
with less than primary education; and in Model 2 only BRG
class V remained associated with poor mental health (Table 5).
As an alternative to Wrights detailed social class structure,
we explored social class relations of property, organizational,
and credential assets separately from each other. Results indicate
that poor health status is associated with supervisor, non-
managerial, semi-skilled, and unskilled class positions among
men and with semi-skilled and unskilled class positions among
women. Poor mental health was also associated with semi-
skilled and unskilled class positions among women. Among
both men and women, credentials are associated with general
health after adjustment for education and BRG. Having a
credentialed occupation is protective of health over and above
the amount of education needed to gain access to that kind of
occupation. That is, the same amount of education would not
protect a persons health that much if that person did not use it
to gain a more advantaged social class.
Discussion
The findings from the Barcelona 2000 HIS support several of
our hypotheses with regard to the relationships between social
class and health. Among men, managers and supervisors with
high credentials had better self-perceived health than men in
other class positions, most notably semi-skilled and unskilled
workers, semi-skilled and unskilled supervisors, petit bourgeois,
and small employers. Although the findings for women were
consistent with those for men, the managerial (i.e. organ-
izational assets) hypothesis could not be confirmed among
women or for mental health outcomes, except for low-level
supervisors, due to large CI, particularly in multivariate models.
Neither social stratification nor social class are related to mental
health in men. This is commonly found with the GHQ.
3436
It
is important to note that although social class was not associated
with mental health, the stratification measures, including the
education measure, were not associated with mental health
either. On the other hand, among women both measures of
social stratification were associated with mental health.
Among those in capitalist class positions, poor self-perceived
health was rare; however, the small number of representatives
of this class in our sample (46 men and 16 women) reduced the
power of tests involving ownership relations in multivariate
analyses. This problem has also been noted in sociological
surveys.
37
In addition, as shown in a British survey on class
structure,
38
capitalists who participate in general surveys are
more prone to be misclassified than members of other class
positions, and further the wealth and power held by large
SOCIAL STATUS AND HEALTH IN SPAIN 955
Table 4 Bivariate and multivariate associations between poor self-perceived health status, social class, and social stratification. Working men and
women population, 1664 years old. Barcelona 2000
Men Women
Bivariate
a
Model 1 Model 2 Bivariate
a
Model 1 Model 2
Variables OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI
EO Wright classification
Capitalists 2.39 0.49, 11.7 1.99 0.40, 9.86 1.93 0.38, 9.83 1.95 0.37, 10.29 1.60 0.30, 8.60 1.44 0.26, 8.13
Small employers 5.86 2.14, 16.02 4.22 1.50, 11.87 4.39 1.46, 13.2 1.92 0.69, 5.30 1.28 0.43, 3.78 1.09 0.34, 3.42
Petit bourgeoisie 4.49 1.63, 12.36 2.98 1.04, 8.50 2.92 0.95, 8.93 1.41 0.51, 3.93 0.92 0.31, 2.73 0.63 0.20, 2.03
Managers experts and
semi-skilled,
supervisors experts 1 1 1 1 1 1
Supervisors
semi-skilled 6.55 2.33, 18.45 4.97 1.72, 14.37 5.05 1.60, 15.9 1.14 0.30, 4.39 1.04 0.27, 4.04 0.78 0.19, 3.18
Supervisors unskilled 6.64 2.30, 19.19 4.15 1.36, 12.66 4.41 1.33, 14.6 2.74 0.95, 7.91 1.78 0.56, 5.67 1.27 0.37, 4.36
Workers experts 2.20 0.70, 6.87 2.34 0.75, 7.32 2.50 0.79, 7.90 0.63 0.19, 2.07 0.64 0.19, 2.07 0.54 0.16, 1.81
Workers semi-skilled 4.91 1.87, 12.83 3.43 1.27, 9.24 3.41 1.16, 10.0 1.79 0.72, 4.47 1.50 0.59, 3.80 1.11 0.40, 3.08
Workers unskilled 7.63 2.99, 19.47 4.49 1.65, 12.16 4.37 1.47, 13.0 3.29 1.38, 7.82 2.00 0.75, 5.32 1.25 0.43, 3.65
2 log likelihood 1464.8 (P 0.00
b
) 1464.1 (P 0.00
b
) 1459.0 (P 0.00
b
) 1376.2 (P 0.00
b
) 1376.2 (P 0.00
b
) 1376.2 (P 0.00
b
)
British Registrar General classification
I 1 1 1 1
II 0.75 0.35, 1.59 0.55 0.25, 1.21 1.57 0.75, 3.27 1.56 0.72, 3.38
III 2.16 1.34, 3.48 1.19 0.65, 2.18 2.37 1.30, 4.32 1.74 0.83, 3.68
IV 3.40 2.15, 5.39 1.96 1.07, 3.56 4.53 2.49, 8.24 3.39 1.58, 7.28
V 4.11 2.15, 7.86 2.19 1.00, 4.76 6.19 3.31, 11.59 4.54 2.02, 10.2
2 log likelihood 1450.3 (P 0.00
b
) 1437.1 (P 0.00
c
) 1359.6 (P 0.00
b
) 1345.9 (P 0.00
c
)
Educational level
University 1 1 1 1
Secondary 1.88 1.24, 2.84 1.28 0.79, 2.07 2.00 1.34, 2.98 1.29 0.75, 2.22
Primary 3.25 2.23, 4.74 2.15 1.35, 3.40 2.65 1.81, 3.87 1.68 0.98, 2.88
Less than primary 3.82 2.22, 6.60 2.58 1.41, 4.73 6.78 3.99, 11.54 4.43 2.31, 8.51
2 log likelihood 1462.1 (P 0.00
b
) 1445.1 (P 0.00
c
) 1368.5 (P 0.00
b
) 1350.3 (P 0.00
c
)
a
Different regressions models for each variable.
b
P-value of the model.
c
P-value of adding the new variable (BRG classification or educational level) to the model after inclusion of other measures.
Age was also included in the models.

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

employers makes them less likely to be reached or to be
motivated to participate in surveys. This may be a limitation of
contemporary survey research that could be overcome with
qualitative research (e.g. ref. 39). The poor health of the petit
bourgeois in our sample could reflect the competition (and high
rates of business failure) that this social class typically experi-
ences, especially in the 21st century marketplace dominated by
large corporations.
1,8,10
Similar results were described by
Benach et al.
40
Using the 3rd European Survey on Working
Conditions, they found that in the European Union (EU) small
employers were at greater risk of reporting high levels of stress
and fatigue and low dissatisfaction and absenteeism.
Interestingly, in our study men in low-level supervisory class
positions (i.e. unskilled supervisors) showed a higher rate of
poor mental health than semi-skilled and unskilled workers.
This is consistent with the notion that contradictory class
relations are mentally hazardous. Multivariate results showing
that unskilled supervisors, but not semi-skilled or unskilled
workers, were more likely to present poor mental health
than managers and expert supervisors, are also consistent
with this. Low-level supervisors are the de facto management
to workers, while simultaneously occupying the position of
workers in relation to upper management, and they are in
conflict with both.
9
These findings are consistent with the
results from a survey conducted in Baltimore (USA), in the
mid 1990s.
5
As expected, experts were found to enjoy better health than
non-experts. The health consequences of the skill-credentials
dimension may be crucial for individuals occupying dual class
positions. Scarce credentials (i.e. expertise) confer a notable
health benefit to low-level supervisors. However, because Wrights
indicators of skills/credentials are similar to occupational strati-
fication, it is unclear whether the skill/credential measure is
actually a measure of social stratification or a measure of social
class proper.
The finding that credentialed occupations are protective of
health over and above the amount of education needed to gain
access to that occupation is consistent with a materialist rather
than a psychosocial interpretation.
16
In addition to the limitations of cross-sectional data for draw-
ing causal inferences, our survey was restricted to the employed
population and thus we were not able to ascertain the social class
positions of those working outside the labour market, such as the
mediated class positions of family members (dependent elderly,
children, housewives devoted to unpaid household labour), or
class trajectory positions (students). This is important mainly for
women because in Spain the participation of women in the
labour market is lower than in other countries of the EU (in
1998 womens activity rate was 37%, whereas the mean of the
EU was 46%).
41
Although the survey was large, some lack of
power was noted in using Wrights class scheme given the number
of categories it contains. In future research involving general
956 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 5 Bivariate and multivariate associations between poor mental health status, social class and social stratification. Working men and women
population, 1664 years old. Barcelona 2000
Men Women
Bivariate
a
Model 1 Model 2 Bivariate
a
Model 1 Model 2
Variables OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI
EO Wright classification
Capitalists 1.62 0.48, 5.45 1.54 0.45, 5.26 1.69 0.49, 5.87 1.52 0.25, 9.21 1.34 0.22, 8.25 1.60 0.25, 9.99
Small employers 1.87 0.85, 4.10 1.79 0.79, 4.03 1.98 0.83, 4.73 2.74 1.04, 7.24 2.16 0.78, 5.95 2.51 0.88, 7.17
Petit bourgeoisie 1.77 0.80, 3.88 1.69 0.74, 3.86 1.71 0.71, 4.14 2.08 0.78, 5.57 1.72 0.62, 4.79 1.69 0.58, 4.85
Managers experts and
semi-skilled,
supervisors experts 1 1 1 1 1 1
Supervisors
semi-skilled 1.77 0.76, 4.10 1.67 0.70, 3.97 1.82 0.71, 4.65 2.08 0.66, 6.55 1.97 0.62, 6.21 1.98 0.61, 6.41
Supervisors unskilled 2.99 1.35, 6.65 2.80 1.18, 6.65 3.28 1.27, 8.51 2.95 1.08, 8.03 2.25 0.77, 6.58 2.55 0.84, 7.78
Workers experts 1.30 0.57, 2.97 1.32 0.58, 3.03 1.17 0.50, 2.75 1.48 0.56, 3.90 1.52 0.58, 4.01 1.35 0.51, 3.62
Workers semi-skilled 1.90 0.96, 3.77 1.82 0.89, 3.73 1.89 0.85, 4.19 2.65 1.12, 6.28 2.44 1.02, 5.82 2.40 0.95, 6.02
Workers unskilled 1.79 0.92, 3.49 1.69 0.80, 3.50 1.92 0.83, 4.44 2.57 1.11, 5.93 1.91 0.76, 4.79 1.94 0.73, 5.14
2 log likelihood 1394.3 (P = 0.42
b
) 1393.0(P = 0.42
b
) 1386.2 (P = 0.42
b
) 1563.4 (P = 0.21
b
) 1563.4 (P = 0.21
b
) 1563.4 (P = 0.21
b
)
British Registrar General classification
I 1 1 1 1
II 1.64 0.96, 2.79 1.37 0.76, 2.46 1.50 0.88, 2.57 1.33 0.76, 2.35
III 1.40 0.89, 2.22 0.91 0.50, 1.65 1.29 0.82, 2.04 1.02 0.58, 1.79
IV 1.40 0.89, 2.20 0.96 0.53, 1.73 1.91 1.21, 3.02 1.55 0.86, 2.81
V 1.80 0.91, 3.57 1.18 0.52, 2.66 2.35 1.39, 3.97 2.01 1.03, 3.93
2 log likelihood 1391.2 (P = 0.46
b
) 1383.4 (P = 0.58
c
) 1565.6 (P 0.00
b
) 1550.8 (P = 0.01
c
)
Educational level
University 1 1 1 1
Secondary 1.44 1.00, 2.07 1.17 0.75, 1.83 1.56 1.13, 2.15 1.46 0.93, 2.28
Primary 1.27 0.88, 1.84 1.03 0.65, 1.64 1.28 0.91, 1.79 1.22 0.76, 1.95
Less than primary 1.15 0.57, 2.34 0.94 0.44, 2.01 2.21 1.27, 3.84 2.10 1.10, 4.00
2 log likelihood 1400.2 (P = 0.40
b
) 1392.0 (P = 0.83
c
) 1571.1 (P = 0.02
b
) 1557.0 (P = 0.09
c
)
a
Different regressions models for each variable.
b
P-value of the model.
c
P-value of adding the new variable (BRG classification or educational level) to the model after inclusion of the other measures.
Age was also included in the models.

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

SOCIAL STATUS AND HEALTH IN SPAIN 957
population samples, the number of categories in Wrights class
framework should perhaps be reduced. Additionally, in studies of
social class inequalities in health, specific class positions (managers,
capitalists) may need to be over-sampled, just as ethnic and
racial minorities must be over-sampled.
Among men, neither the occupation-based BRG nor education
seems to be a better predictor of self-perceived health than social
class. Among women, we did not obtain strong associations. How-
ever, non-measured aspects of gender (exposure to worse working
conditions and lack of access to labour markets; household labour
and social networks
9,14,42
) could account for these results.
Our findings add to the literature on comparative indicators of
social inequalities and highlight the importance of control over
material resources,
4347
or in Wrights terms, control over organ-
izational assets. Our results confirm recent studies
57
in which
social class, understood as a social relation of ownership or control
over productive assets, explains some aspects of the variation in
health outcomes, while social stratification explains others.
We can draw several conclusions from this study. Our find-
ings suggest that surveys in social epidemiology could benefit
from over-sampling large employers in order to assess the
health impact of capitalist class positions, which are poorly
represented in general population samples.
38,48
The poor
mental health found among low-level supervisors, replicating a
previous study, suggests that inquiry into the mental health
effects of contradictory class positions may be a fruitful venue
for future research. Furthermore, our study findings indicate that
control over organizational assets, as captured by the power to
hire and fire labour and decision-making power over company
policy, may be an important determinant of social inequalities
in health. Thus, our findings highlight the potential health
consequences of social class positions defined by power
relations within the labour process. They also confirm that
social class taps into parts of the social variation in health that
are not captured by conventional measures of social stratification
and education.
Acknowledgements
This study was supported with funds from the Municipal
Institute of Public Health of Barcelona, Spain. The authors want
to thank Wylbur Hadden for his helpful comments.
References
1
Wright EO. Class Counts: Comparative Studies in Class Analysis.
Cambridge: Cambridge University Press, 2000.
2
Lynch J, Kaplan G. Socioeconomic position. In: Berkman Lf, Kawachi I
(eds). Social Epidemiology. New York: Oxford University Press, 2000.
3
Muntaner C, Eaton WW, Diala CC. Socioeconomic Inequalities in
mental health: a review of concepts and underlying assumptions.
Health 2000;47:204353.
4
Krieger N, Williams DR, Moss N. Measuring social class in US public
health research: concepts, methodologies and guidelines. Ann Rev
Public Health 1997;18:34178.
5
Muntaner C, Eaton WW, Diala CC, Kessler RC, Sorlie PD. Social class,
assets, organizational control and the prevalence common groups of
psychiatric disorders. Soc Sci Med 1998;47:204353.
6
Wohlfarth T. Socioeconomic inequality and psychopathology: are
socioeconomic status and social class interchangeable? Soc Sci Med
1997;45:399410.
7
Wohlfarth T, van den Brink W. Social class and substance use
disorders: the value of social class as distinct from socioeconomic
status. Soc Sci Med 1998;47:5158.
8
Wright EO, Singleman J. Proletarization in the changing American
class structure. Am J Sociol 1982;88:176209.
9
Wright EO. Classes. London: Verso, 1985.
10
Steinmetz G, Wright EO. The fall and rise of the Petty bourgeoisie:
changing patterns of employment in the postwar United States. Am J
Sociol 1989;94:9731018.
11
Wright EO, Cho D. The relative permeability of class boundaries to
cross class friendships: a comparative study of the United States,
Canada, Sweden, and Norway. Am Soc Rev 1992;57:85102.
12
Muntaner C, Wolyniec P, McGrath J, Pulver AE. Psychotic inpatients
social class and their first admission to state or private psychiatric
Baltimore hospitals. Am J Public Health 1994;84:28789.
13
Western M, Wright EO. The permeability of class boundaries to
intergenerational mobility among men in the United States, Canada,
Norway and Sweden. Am Soc Rev 1994;59:60629.
14
Wright EO, Baxter J, Birkelund GE. The gender gap in workplace
authority: a cross-national study. Am Soc Rev 1995;60:40735.
15
Wright EO. Race, class, and income inequality. Am J Sociol 1978;
83:136897.
16
Lynch J, Due P, Muntaner C, Smith GD. Social capital, is it a good
investment strategy for public health? J Epidemiol Community Health
2000;54:40408.
17
Keister L. Wealth in America. Trends in Wealth Inequality. Cambridge:
Cambridge University Press, 2000.
18
Turner RJ, Roszell P. Psychosocial resources and the stress process. In:
WR Avison, IH Gotlib. Stress and Mental Health. New York: Plenum,
1994, pp. 179212.
19
Wolff EN. Top Heavy: a Study of Wealth Inequity in America. New York:
20th Century Fund, 1996.
20
Spilerman S. Wealth and stratification processes. Ann Rev Sociol
2000;26:497524.
21
Keister LA. Financial markets, money, and banking. Ann Rev Sociol
2002;28:3961.
22
Halaby CN, Weakliem DL. Ownership and authority in the earnings
functions. Am Soc Rev 1993;58:1630.
23
Karasek K, Theorell T. Healthy Work: Stress, Productivity and the
Reconstruction of Working Life. NY: Basic Books, 1990.
24
Marmot MG, Davey Smith G, Stansfeld et al. Health inequalities
among British civil servants: the Whitehall II study. Lancet 1991;
337:138793.
25
Idler EL, Benyamini Y. Self-rated health and mortality: a review of
twenty-seven community studies. J Health Soc Behav 1997;38:2137.
26
Goldberg D. The Detection of Psychiatric Illness by Questionnaire. London:
Oxford University Press, 1972.
27
Kaplan G, Baron-Epel O. What lies behind the subjective evaluation
of health status? Soc Sci Med (In press).
28
Hennessy CH, Moriarty DG, Zack MM, Scherr PA, Brackbill R.
Measuring health-related quality of life for public health surveillance.
Public Health Rep 1994;109:66572.
29
Borrell C, Rue M, Pasarn MI et al. Trends in social class inequalities in
health status, health-related behaviors, and health services utilization
in a southern European urban area (19831994). Prev Med 2000;
31:691701.
30
Ajuntament de Barcelona. Encuesta de Poblacin Activa del Instituto
Nacional de Estadstica, 2001.
31
Carabaa J, de Francisco A (eds). Teoras Contemporneas De Las Clases
Sociales. Madrid: Editorial Pablo Iglesias, 1994.
32
Grupo SEE y Grupo SEMfYC. Una propuesta de medida de la clase
social. Atencin Primaria 2000;25:35063.

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Muntaner et al.
1
have taken on one of the major issues for social
epidemiology today. For many years now, we have seen study
after study showing relationships between health and some
measure or other of socioeconomic position and circumstances.
But, with honourable exceptions, studies have rarely faced up
directly to the problems involved in the conceptualization and
measurement of socioeconomic position (SEP).
26
There is
seldom any account of how it has been decided to use one or
other method. Studies are then compared with little consider-
ation for the fact that some use indicators of income, others use
indicators of prestige, and others use education.
7,8
Even to
make the distinction between class and prestige is likely to
sound rather strange to many readers.
As Muntaner et al. point out, the most commonly used meas-
ures of SEP have been indicators of position in a system of
stratification. The term stratification is used in sociology to refer
to social hierarchies in which individuals or groups can be
arranged along a ranked order of some attribute. Income is one
such. A more popular one is social status or prestige.
9
In the US,
there have been several large scale surveys that ask respondents
to rank large numbers of occupations
10
in terms of prestige.
However, these lists cannot be exhaustive. The most widely used
measures in the US therefore extrapolate prestige for all other
occupations on the basis of income and education levels.
1012
This is a perfectly consistent procedure, because prestige,
income, and education are all ordered from high to low. There
may be problems in mixing the three up together, rather than
trying to estimate their relationships to health separately, but
that is a different story.
13
The measurement of social class, if we use this term in the
sociological sense, is quite different. Class, in contrast to stratifi-
cation, indicates the employment relations and conditions of
each occupation. The criteria used to allocate occupations into
classes vary somewhat between the two major systems presently
in widespread use: the Goldthorpe schema (and its develop-
ments) and the Wright schema.
1417
Muntaner et al. describe
clearly the classificatory principals of the Wright schema, which
they used in their paper. Official statistics in Great Britain now
958 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
33
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, John
Wiley and Sons Inc, 1989.
34
Siahpush M, Singh GK. A multivariate analysis of the association
between social class of origin and current social class with self-rated
general health and psychological health among 16-year-old
Australians. Aust N Z J Med 2000;30:65359.
35
Stansfeld SA, Marmot MG. Social class and minor psychiatric disorder
in British Civil Servants: a validated screening survey using the
General Health Questionnaire. Psychol Med 1992;22:73949.
36
Goldberg DP, Rickels K, Downing R, Hesbacher P. A comparison of
two psychiatric screening tests. Br J Psychiatry 1976;129:6167.
37
Kohn M, Schooler, C. Work and Personality. An Inquiry into the Impact of
Social Stratification. Norwood, New Jersey, Ablex, 1983.
38
Marshall G, Rose D, Vogler C, Newby H. Social Class in Modern Britain.
London, Hutchinson, 1988.
39
Sennet R, Cobb J. The Hidden Injuries Of Class. New York: Vintage, 1972.
40
Benach J, Gimeno D, Benavides FG. Types of Employment and Health in
the European Union. European Foundation for the Improvement of Living
and Working Conditions. Luxembourg: Office of Official Publication
of the European Communities, 2002.
41
Subdireccin General de Proceso de Datos del Ministerio de Trabajo y
Asuntos Sociales. Anuario de Estadsticas Laborales y Sociales 1998.
Madrid: Ministerio de Trabajo y Asuntos Sociales 1999.
42
Sacker A, Firth D, Fitzpatrick R, Lynch K, Bartley M. Comparing
health inequality in men and women: prospective study of mortality
198696. BMJ 2000;320:130307.
43
Davey Smith G, Hart C et al. Education and occupational social class:
which is the more important indicator of mortality risk? J Epidemiol
Community Health 1998;52:15360.
44
Prandy K. Class, stratification, and inequalities in health. Sociol Health
Illness 1999;21:46685.
45
Chandola T, Jenkinson C. The new UK National Statistics Socio-
Economic Classification (NS-SEC); investigating social class
differences in self-reported health status. J Public Health Med
2000;22:18290.
46
Chandola T. Social class differences in mortality using the new UK
National Statistics Socio-Economic Classification. Soc Sci Med
2000;50:64149.
47
Geyer S, Peter R. Income, occupational position, qualification and
health inequalitiescompeting risks? (comparing indicators of social
status). J Epidemiol Community Health 2000;54:299305.
48
Muntaner C, Parsons PE. Income, stratification, class and health
insurance. Int J Health Services 1996;26:65571.
IJE vol.32 no.6 International Epidemiological Association 2003; all rights reserved. International Journal of Epidemiology 2003;32:958960
DOI: 10.1093/ije/dyg232
Commentary: Relating social structure
and health
M Bartley
Department of Epidemiology and Public Health, University College London,
119 Torrington Place, London WC1E 6BT, UK. E-mail: m.bartley@public-
health.ucl.ac.uk

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

use a development of the Goldthorpe schema, the National
Statistics Socio-Economic Classification (NS-SEC).
18,19
What
these have in common are a concern with power and control at
the workplace: control over ones own working day and work
career and power over the work of others. Thus, the conceptual
basis for social class is totally different to that underlying
stratification. Most importantly, class is an inherently relational
concept. It is not defined according to an order or hierarchy, but
according to relations of power and control.
According to Wright, power or authority are organisational
assets that allow some workers to benefit from the abilities and
energies of other workers.
17
The hypothetical pathway linking
class (as opposed to prestige) to health is that some members of
a work organization are expending less energy and effort and
getting more (pay, promotions, job security, etc.) in return,
while others are getting less for more effort. So the less
powerful are at greater risk of running down their stocks of
energy and ending up in some kind of physical or psychological
health deficit. French industrial sociologists called this lusure
de travailthe usury of work. At the most obvious level, the
manager sits in an office while the routine workers are exposed
to all the dangers of heavy loads, dusts, chemical hazards and
the like. But we are all familiar with other situations: the
academic supervisor who takes credit for the work of a graduate
student or junior researcher; the manager who takes the credit
for the efficiency of a secretary. The same process occurs in
every organization where power is unequal. So the use of the
Wright schema in health research, as Muntaner et al. point out
Provides an explicit relational mechanism that explains how
economic inequalities are generated and how they may affect
health.
The authors have used, as their contrast to the Wright class
measure, not one of the American socioeconomic status
measures but a Spanish derivation of the British Registrar-
Generals classification (RGSC). The RGSC has been described
variously as an indicator of skill level or of general standing in
the community (which sounds like prestige). But there was
never any system behind the measure, most occupations were
given the same class as in the previous decennium unless a
member of the decision making group happened to have
enough knowledge of any specific one to raise questions. The
story used to be told about one of the civil servants deciding on
the class position of a newly emerging occupation on the basis
of whether he would like his daughter to marry one. This is, of
course, a perfectly valid method for the allocation of caste
membership in ethnographic studies. Given that the measure
has been widely used in Spanish epidemiological studies, and
that it is used here as an indicator of the RG classification to
indicate a prestige dimension of stratification, it seems a
reasonable choice.
Muntaner et al. directly address the extent to which the use
of a specific class measure actually tells us any more than we
could learn from a measure of income or of prestige by
considering the health of people in some of Wrights
contradictory class locations. Supervisors have higher incomes
and prestige than non-supervisory workers, so on this basis
their health should be better. However, taking account of the
relationships between supervisors and other workers might lead
to a different expectation. There are contradictory pressures on
supervisors, in that they are subject to the authority of those
above them but also suffer the antagonism of those below. Sure
enough, the worst mental (not physical) health in the authors
analysis was found amongst unskilled supervisors; the group
thought most likely to have responsibility without very much
power. This effect seems to have been independent of any
association between Wright class and education or the Spanish
version of the RG schema.
An important point to emerge from the results is the dis-
tinction between expertise and education. In fact, the relative
weakness of education itself as a predictor of health is striking.
After adjustment for educational level, experts of various kinds
seem to have relatively low risk of ill-health, both mental and
physical. The paper is therefore showing us that it is not edu-
cation per se that favourably influences health, but the access
education gives to more dominant (or less-dominated) social
positions. Of course, any examination of the relationships of
class or education to health needs to take account of the relevant
national labour market. There may be very large differences
between the experience of being the owner of a small or medium
sized company in Spain, the UK, and the US. But in many
countries, the employment situation of someone designated as
an expert within a large organization, say a university or gov-
ernment bureaucracy, may be little different to an extended
vacation. The owner/manager of a company employing
20 people (who counts as a capitalist by the Wright criteria), on
the other hand, may be consigned to near-slavery by
comparison.
Indiscriminate use of measures of social position with multiple
theoretical bases (or none at all) has hindered progress from the
description of health inequality towards its explanation. It has
been all too easy to slip into the kind of lazy thinking that
proceeds: high social class equals general superiority in lots
of ways equals better health, what do you expect?. Relational
social class measures may show weaker associations to some
health outcomes than prestige or income. But then we will at
least be able to eliminate some potential causal pathways and
concentrate on those that are better supported by the data.
References
1
Muntaner C. The associations of social class and social stratification
with patterns of general and mental health in a Spanish population.
Int J Epidemiol 2003;32:95058.
2
Krieger N, Moss N. Accounting for the publics health: an introduction
to selected papers from a U.S. conference on measuring social
inequalities in health. Int J Health Serv 1996;26:38390.
3
Muntaner C, Parsons PE. Income, social-stratification, class, and
private health-insurancea study of the Baltimore metropolitan-area.
Int J Health Serv 1996;26:65571.
4
Krieger N, Williams DR, Moss NE. Measuring social class in US public
health research: Concepts, methodologies, and guidelines. Annu Rev
Public Health 1997;18:34178.
5
Wohlfarth T. Socioeconomic inequality and psychopathology: are
socioeconomic status and social class interchangeable? Soc Sci Med
1997;45:399410.
6
Wohlfarth T, van den Brink W. Social class and substance use
disorders: the value of social class as distinct from socioeconomic
status. Soc Sci Med 1998;47:5158.
7
Bartley M, Sacker A, Firth D, Fitzpatrick R. Understanding social
variation in cardiovascular risk factors in women and men: the
SOCIAL STATUS AND HEALTH IN SPAIN 959

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

advantage of theoretically based measures. Soc Sci Med 1999;
49:83145.
8
Duncan GJ, Daly MC, McDonough P, Williams DR. Optimal indicators
of socioeconomic status for health research. Am J Public Health 2002;
92:115157.
9
Hodge RW. The measurement of occupational status. Soc Sci Res
1981;10:396415.
10
Duncan OD. A socioeconomic index for all occupations. In: Reiss Jr
AJ. Occupations and Social Status. New York: Free Press, 1961,
pp. 10938.
11
Nam CB, Terrie WE. Measurement of socioeconomic status from
United States Census data. In: Rossi PH, Nock SL (eds). Measuring
Social Judgements: The Factorial Survey Approach. Beverley Hills: Sage,
1982, pp. 95118.
12
Oakes JM, Rossi PH. The measurement of SES in health research:
current practice and steps toward a new approach. Soc Sci Med
2003;56:76984.
13
Hauser RM, Warren JR. Socioeconomic Indexes for Occupations: A Review,
Update and Critique. Madison, Wisconsin: Centre for Demography and
Ecology, University of Wisconsin-Madison, 1996.
14
Erikson R, Goldthorpe JH. The Constant Flux. Oxford: Clarendon, 1992.
15
Evans G, Mills C. In search of the wage-labour/service contract: new
evidence on the validity of the Goldthorpe class schema. Br J Sociol
2000;51:64161.
16
Goldthorpe JH. The Goldthorpe class schema: some observations on
conceptual and operational issues in relation to the ESRC review of
government social classification. In: Rose D, OReilly K (eds). Con-
structing Classes: Toward a New Social Classification for the UK. Swindon:
ONS/ESRC, 1997, pp. 4048.
17
Wright EO. Class Counts. Cambridge: Cambridge University Press,
1997.
18
Rose D, Pevalin DJ. The National Statistics Socio-economic Classification:
Unifying official and sociological approaches to the conceptualisation and
measurement of social class. ISER working paper no. 20014. Colchester:
ISER, University of Essex, 2001.
19
Rose D, Pevalin DJ. The NS-SEC explained. In: Rose D, Pevalin DJ
(eds). A Researchers Guide to the National Statistics Socio-economic
Classification. London: Sage, 2003, pp. 2844.
960 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

a
t

T
h
e

R
e
f
e
r
e
n
c
e

S
h
e
l
f

o
n

A
p
r
i
l

8
,

2
0
1
1
i
j
e
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m