1. health and fitness

ARTICLE IN PRESS
Social Science & Medicine 65 (2007) 1965–1978
www.elsevier.com/locate/socscimed
The problems of relative deprivation: Why some societies do
better than others
Richard G. Wilkinsona,, Kate E. Pickettb
a
Division of Epidemiology and Public Health, University of Nottingham, Nottingham NG7 2UH, UK
b
Department of Health Sciences, University of York, UK
Available online 5 July 2007
Abstract
In this paper, we present evidence which suggests that key processes of social status differentiation, affecting health and
numerous other social outcomes, take place at the societal level. Understanding them seems likely to involve analyses and
comparisons of whole societies.
Using income inequality as an indicator and determinant of the scale of socioeconomic stratification in a society, we
show that many problems associated with relative deprivation are more prevalent in more unequal societies. We summarise
previously published evidence suggesting that this may be true of morbidity and mortality, obesity, teenage birth rates,
mental illness, homicide, low trust, low social capital, hostility, and racism. To these we add new analyses which suggest
that this is also true of poor educational performance among school children, the proportion of the population imprisoned,
drug overdose mortality and low social mobility.
That ill health and a wide range of other social problems associated with social status within societies are also more common
in more unequal societies, may imply that income inequality is central to the creation of the apparently deep-seated social
problems associated with poverty, relative deprivation or low social status. We suggest that the degree of material inequality in a
society may not only be central to the social forces involved in national patterns of social stratification, but also that many of the
problems related to low social status may be amenable to changes in income distribution.
If the prevalence of these problems varies so much from society to society according to differences in income
distribution, it suggests that the familiar social gradients in health and other outcomes are unlikely to result from social
mobility sorting people merely by prior characteristics. Instead, the picture suggests that their frequency in a population is
affected by the scale of social stratification that differs substantially from one society to another.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Income inequality; Socioeconomic status; Health; Education; Relative deprivation; Prisons
Introduction
A typical approach to examining contextual area
effects on health is to start by controlling out the
Corresponding author.
E-mail addresses: richard.wilkinson@nottingham.ac.uk
(R.G. Wilkinson), KP6@York.ac.uk (K.E. Pickett).
0277-9536/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2007.05.041
compositional effects of the socioeconomic characteristics of the population in those areas (DiezRoux, 1998; Merlo, 2003; Pickett & Pearl, 2001).
Because individual characteristics usually have the
most powerful influence on the local health profile,
researchers adjust for them in order to see whether
there are residual positive or negative health effects
associated with features of the area itself.
ARTICLE IN PRESS
1966
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
At the local area level the proportion of people
with various socioeconomic characteristics may be
primarily a distributional issue—how does an area
become a deprived area, inhabited by a disproportionate share of poorer, less well-educated people,
while a neighbouring area attracts better off people
(Macintyre, Maciver, & Sooman, 1993; Tunstall,
Shaw, & Dorling, 2004)? However, on another level
this is not a question about the distribution in
physical space of people with given characteristics,
it is instead about the social forces which create
those characteristics in the first place. What
determines the proportions of people in the wider
society belonging to different social classes, in
different income groups, with different levels of
educational qualifications? The answer to a question
like that may seem to depend on the outcome
of hundreds of different processes covering
every aspect of poverty and wealth creation:
educational policies and teaching methods, social
mobility, cycles of deprivation, the durability
of class cultures—to name but a few, all involving
the complexities and minutiae of interactions
between individuals, their social environments,
and the wider society. However, in this paper
we will show that a wide range of problems
associated with relative deprivation (including ill
health, teenage births, violence, low trust, the
educational performance of school children, imprisonment, drug abuse, and obesity), are all strongly
related to one factor—societal measures of income
distribution.
Rather than being left with the infinite complexities of different determinants of people’s standard
of health, education, propensities to violence, risks
of teenage pregnancy, imprisonment, etc., and
trying to formulate separate policies which might
have an impact on each of these, it may be that there
are also some simpler patterns. If the distribution of
each of these health and social problems is related to
relative deprivation within societies, and also all
tend to be more common in more unequal societies,
then perhaps this tells us something fundamental
about the impact of the processes of social
differentiation within the population.
In this paper, we consider income inequality as
both an indicator and a determinant of the scale of
social stratification in a society. The range of
outcomes which we shall show are statistically
related to income distribution suggests that income
inequality is related to deep-seated processes of
social differentiation. Rather than thinking of
populations as made up of basically similar people
who—by luck or judgement—have become attached
to different incomes and make up societies with
different levels of income inequality, we suggest that
the social processes related to income distribution
are involved in the deeper ways our personal and
class characteristics are constituted (Bourdieu,
1984). As Williams (1995) argues, in discussing
Bourdieu’s complex concept of ‘‘habitus’’ it ‘‘is the
(class-related) habitus which y determine(s) not
only lifestyles and the chances of successybut also
class-related inequalities in health and illness’’ and,
as we would argue, also determines class-related
inequalities in many other outcomes. The social
processes which become structured round income
distribution probably also include many of the early
childhood influences on social and cognitive development which seem to affect both health and social
mobility and are important in social class differentiation (Ben-Shlomo & Kuh, 2002).
Our theory is that processes of social status
differentiation, including whether a society has a
more or less hierarchical class structure, are
intimately related to the scale of income distribution. For several reasons we believe, along with
others, that these processes are structured primarily
at the national level. As Taylor and Flint (2000)
state, ‘‘classes have most commonly defined themselves on a state-by-state basis’’. Our thinking is not
only based on the fact that the distribution of
income reflects market incomes (earned and unearned) from the national economy, plus the effects
of varying degrees of redistribution resulting from
national systems of taxes and benefits; it is also
informed by our recent review of the literature on
income inequality and health (Wilkinson & Pickett,
2006). We found, as had been noted in previous
studies and commentaries (Franzini, Ribble, &
Spears, 2001; Wilkinson, 1997), that population
health is most reliably related to income distribution
when income differences are measured across
nation-states and other large geo-political units.
Indeed, the evidence suggested a graded relationship
such that small area studies in parishes, neighbourhoods and counties showed either weak or nonexistent relationships; studies of states, regions and
cities tended to show stronger, more consistent
relationships; and studies comparing nation-states
showed the strongest and most consistent evidence.
This observation is given additional weight by the
fact that the same pattern was independently
reported in an earlier review of studies looking at
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
the relationship between income inequality and
violence (Hsieh & Pugh, 1993).
We realise that there has been considerable
debate within the disciplines of economic and
political geography concerning the spatial scales at
which social processes are structured, and the
complex ways in which systems operating at
different scales interact. However, as Brenner
(2001, p. 606) points out ‘‘Whether or not the scalar
structuration of a given social process generates
sociologically or politically significant outcomes is
an empirical question that can only be resolved
through context-specific inquiries’’ (Brenner, 2001).
We are sensitive to Marston’s point that ‘‘contemporary writing about scale in human geography has
failed to comprehend the real complexity behind the
social construction of scale’’ (Marston, 2000) and
her argument that the emphasis has too often been
on the functional agency of ‘‘the international
economy’’ or to ‘‘national social formations’’, while
‘‘other social practices are cordoned off in their
respective localities’’ ignoring the fact ‘‘that even the
most privileged social actorsyare no less (locally)
situated than the workers they seek to command’’
(Marston, Jones, & Woodward, 2005). Nevertheless, the role of nationally constituted social
differentiation in relation to health and social
outcomes remains an open question, and one that
can be empirically examined.
Income inequalities in large areas can of course be
decomposed into income inequalities within and
between their smaller constituent areas (Lobmayer
& Wilkinson, 2002). And of course social comparisons with neighbours may sometimes have detectable effects on health. Nevertheless, as Ballas asks,
‘‘do (people) compare themselves to ‘‘peer groups’’
in their neighbourhood, city, region, country or
possibly to diaspora groups in other countries or
with people of whom they know little? There are
many other kinds of non-geographical groupsyto
which we may compare ourselves and with whom
we consider ourselves to be of a similar social
standing. It is far from clear how reference groups
are constituted’’ (Ballas, Dorling, & Shaw, 2007).
As we have argued previously, the health of people
in a deprived neighbourhood is worse not because
of inequalities within that neighbourhood, but
because they are deprived in relation to the wider
society. To give a particularly dramatic example,
consider that in 1996, black American men had a
median income of $26,522 and an average life
expectancy of only 66.1 years. In comparison, men
1967
in Costa Rica had a mean income (at purchasing
power parity) of only $6410, yet their average life
expectancy was 75 years (Marmot & Wilkinson,
2001). To call any local level of income an effect of
‘‘absolute’’ income (or education or deprivation),
and to assume that its relation to health is
independent of the wider context is to forget that
poor areas are poor in relation to the wider society.
Rather than ignoring the fabric of people’s lives
which, as Marston (2000) pointed out are always
locally situated, we are suggesting that social classes
are constituted in relation to each other partly
through what may look like action at a distance—
through the effects of the population class structure
outside one’s immediate locality.
We start our empirical investigation by summarising previously published evidence suggesting that
the societal scale of income inequality is related to
morbidity and mortality, obesity, teenage birth
rates, mental illness, homicide, low trust, low social
capital, hostility, and racism. We then go on to test
new hypotheses, that poor educational performance
among school children, the proportion of the
population imprisoned, drug overdose mortality
and low social mobility are also related to greater
income inequality. We would emphasise that the
issue throughout is not that greater income inequality means simply greater inequality in outcomes
localised within societies, but that greater income
inequality is associated with a higher prevalence of
ill health and social problems in a society as a
whole, regardless of its social distribution.
Recent evidence linking inequality to social outcomes
Health
Morbidity and mortality
In a recent review of 168 analyses of the relationship
between income inequality and population health, we
found that a large majority of studies reported that
more egalitarian societies tend to be healthier
(Wilkinson & Pickett, 2006). Studies of small
areas—such as parishes and census tracts—were the
only major exceptions to this pattern. We found 104
studies of health in which income inequality was
measured across whole nations, states, regions or
cities—areas large enough for income inequality to be
indicative of the overall scale of social differentiation
and social hierarchy in those societies (Wilkinson &
Pickett, 2006). After adjustment for various control
variables (including ones which could be either
ARTICLE IN PRESS
1968
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
mediating or confounding variables), 81 of the 104
studies (78%) found all or some of the health variables
they measured were significantly related to inequality.
Before adjustment the proportion supporting this
relationship was higher still. As well as a large number
of international comparisons of developed and developing countries, evidence confirming this pattern also
came from studies of regions, states, and cities in a
number of different countries including Canada,
Chile, China, Ecuador, Italy, Russia, Taiwan, UK,
and USA. In contrast, studies that measured inequality and health in smaller areas (counties, tracts, and
parishes) produced more equivocal results.
Obesity
In a study of obesity rates (BMI4in 30) 21 of the
richest countries we reported that rates were higher
in more unequal societies (Pickett, Kelly, Brunner,
Lobstein, & Wilkinson, 2005). These relationships
were statistically significant for obesity among both
men and women, but noticeably stronger among
women. The same study also showed that greater
inequality was associated with higher total calorie
intake. The relation between obesity and inequality
was attenuated, but remained significant, even after
adjusting for calorie consumption.
children. The teenage birth rate was reported to be
closely related to income inequality both internationally among 21 rich countries and among the 50
states of the USA (Gold, Kawachi, Kennedy,
Lynch, & Connell, 2001; Gold, Kennedy,
Connell, & Kawachi, 2002; Pickett, Mookherjee, &
Wilkinson, 2005).
Mental illness
Using surveys of random samples of the population, the World Health Organization (WHO)
recently produced comparable estimates of the
prevalence of mental illness for eight developed
countries—six in Western Europe plus Japan and
the USA (Demyttenaere et al., 2004). We found
statistically significant correlations between income
inequality and the prevalence of both serious and
any mental illness (Pickett, James, & Wilkinson,
2006). We have since confirmed this correlation
(r ¼ 0.79, p ¼ 0.002) in an expanded dataset,
including data from a further WHO survey for
New Zealand, and non-WHO population based
prevalence estimates for Australia, Canada and the
UK (Fig. 1). However, we found no evidence of
such a relation among the 50 states of the USA.
The quality of social relations
Teenage birth rates
Whether for biological or social reasons, teenage
births are often considered a problem with health
and social consequences for both mothers and
Many people have intuited that inequality is
socially divisive and corrosive of human relations.
Writing of the United States in the first half of the
Fig. 1. Prevalence in mental illness in relation to income inequality among rich countries.
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
19th Century, de Tocqueville (2000) emphasised his
belief that the strength of the associational and civic
life to which he drew attention was based (with the
crucial exception of slavery) on what he called the
‘‘equality of conditions’’. Numerous analyses including homicide, trust, social capital, hostility, and
racism, suggest that the quality of social relations in
a society is poorer where there is more inequality
(Wilkinson, 2005).
Homicide
A large body of evidence suggests that there is a
robust relationship between greater inequality and
higher homicide rates. All 24 studies of inequality
and homicide rates in large areas (whole countries,
regions, states or cities) reported significant relationships (Wilkinson & Pickett, 2006). An earlier
review also reported a robust relationship which,
like health, was stronger when the areas measured
were larger rather than smaller (Hsieh & Pugh,
1993).
Trust
There have been a number of analyses of the
relation between inequality and various measures of
the quality of social relations, including trust and
social capital. As summarised elsewhere (Wilkinson,
2005), these results are consistent with the findings
on violence, and suggest that the quality of social
relations is poorer in more unequal societies. An
international analysis of data from 38 countries
(Uslaner, 2002) as well as an analysis among the 50
states of the USA (Kawachi, Kennedy, Lochner, &
Prothrow-Stith, 1997) have shown substantially
lower levels of trust where income differences are
bigger. In the less unequal states only 10% or 15%
felt they could not trust others; this rose to 35%
or 40% in the more unequal states. The differences related to inequality internationally were just
as large.
Social capital
Measures of the strength of associational and
community life have also been reported to be related
to income inequality. Putnam reported a strong
cross-sectional relation between income inequality
and his index of the strength of the ‘‘civic
community’’ in the 20 regions of Italy (Putnam,
1993), and a similar relation between inequality and
his index of social capital among the states of the
USA (Putnam, 2000). Putnam also mentions a
‘‘striking’’ similarity in the trends in inequality and
1969
social capital during the 20th century. Until the late
1960s income differences narrowed and social
capital strengthened, but around 1965–70 both
reversed direction: income differences widened and
social capital weakened.
Hostility and racism
The last indicators that greater inequality is
accompanied by less good social relations come
from US data on hostility scores and racism.
Williams, Feaganes, and Barefoot (1995) measured
hostility scores in random samples of the population
in 10 US cities. The average score for each city
was significantly related to its income inequality
(Wilkinson, 2005). In a separate study, Kennedy,
Kawachi, Lochner, Jones, and Prothrow-Stith
(1997) found that people held more racist attitudes
and beliefs in US states where income differences
were large.
New analyses
In the light of these findings we decided to see if
there were relations between inequality and other
social problems associated with relative deprivation.
The outcomes we were able to look at were limited
by the availability of comparable data but, in
addition to the outcomes discussed above, we now
report analyses of the relationship between income
inequality and the educational performance of
school children, prison populations, drug overdose
mortality, and social mobility.
Data sources and results
To improve comparability, we limited the international analyses to countries among the richest 50
(by Gross National Income per capita at purchasing
power parities) in 2002. To avoid tax havens, we
excluded countries with populations of less than
three million. Income distribution data were available for 24 countries which met these criteria:
Australia, Austria, Belgium, Canada, Denmark,
Finland, France, Germany, Greece, Ireland, Israel,
Italy, Japan, The Netherlands, New Zealand,
Norway, Portugal, Singapore, Slovenia, Spain,
Sweden, Switzerland, the United Kingdom, and
the United States.
Data on income inequality came from the Human
Development Indicators (HDI), 2003; reporting
dates vary slightly from country to country but
are within the period 1992–1998 (United Nations
ARTICLE IN PRESS
1970
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
Development Program, 2003). For Germany we
used an average of the HDI figures given in the
2002–2004 reports as the figure for 2003 was
anomalous. For consistency with our other recent
publications, income inequality was measured as
the ratio of the total annual income received by the
richest 20% of the population to the total annual
income received by the poorest 20% of the
population. This ratio ranged from 3.4 in Japan,
the most equal country, to 9.7 in Singapore, the
most unequal.
To supplement the international analyses, where
data were available we also analysed relationships
with income inequality among the 50 United States
and the District of Columbia (DC). Rather than
calculating our own measure, we used the Gini
coefficient of the inequality of family income for
1999 as provided by the US Census Bureau (2004)
(The Gini coefficient varies between 1, indicating
maximum inequality, and 0, indicating total equality (Allison, 1978)).
Educational performance
To see if the educational performance of school
children was related to inequality internationally,
we used estimates of the combined maths and
literacy scores for 15-year olds taken from the
Programme for International Student Assessment
2003 (OECD Programme for International Student
Assessment, 2004). These data were available for 19
countries and the distribution of educational
performance in relation to income inequality is
shown in Fig. 2. The correlation coefficient was
r ¼ 0.50, p ¼ 0.029.
To check if a similar relationship existed among
the 50 states (and DC) of the USA, we combined
maths and reading performance scores for 8th
graders (about 14-years old) from the US Department of Education, National Center for Education
Statistics for 2003. (US Department of Education,
2004a, 2004b) The scores were significantly lower in
states with wider income differences: r ¼ 0.69,
po0.01. DC was an outlier but the association
remained highly significant when it was excluded.
We also found a statistically significant tendency for
the proportion of children not completing high
school to be greater in more unequal states.
The correlation coefficient was r ¼ 0.66, po0.01.
Imprisonment
Figures on the proportion of the population
imprisoned in different countries come from the
United Nations Survey on Crime Trends and the
Operations of Criminal Justice Systems (United
Nations Crime and Justice Information Network,
2000). Relating these to income inequality we found
a correlation of r ¼ 0.69 po0.01. The data are
shown in Fig. 3. Because the USA is an outlier, we
also checked the association when it was excluded;
the correlation then rose to r ¼ 0.75, po0.01. With
Singapore also excluded the correlation was
r ¼ 0.60, po0.01.
Fig. 2. Educational achievement in relation to income inequality among rich countries.
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
1971
Fig. 3. Imprisonment in relation to income inequality among rich countries.
Fig. 4. Imprisonment in relation to income inequality among the 50 US states and DC, also indicating states with and without the death
penalty.
Again, we looked at the same association among
the 50 states (and DC) of the USA. Figures for
imprisonment in 1997–1998 were taken from the US
Department of Justice (2006), Bureau of Justice
Statistics (US Department of Justice). The correlation coefficient relating imprisonment rates to
income inequality was r ¼ 0.77 (po0.01). DC was
an extreme outlier in this relationship; with much
the highest level of income inequality and an
imprisonment rate of 1566 per 100,000—more than
twice that of Louisiana, the next highest. With DC
excluded, the correlation was attenuated to r ¼ 0.56
but remained statistically significant (p o0.01). The
data (with DC excluded to facilitate scaling) are
shown in Fig. 4, which also shows which states
retain and which have abolished the death penalty.
Abolition appears to be more common in the more
egalitarian states.
Drug overdose mortality
Age-adjusted mortality rates for deaths from
accidental narcotic and hallucinogen poisoning
ARTICLE IN PRESS
1972
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
(ICD-10 Code X42) were taken for US states (and
DC) from the Center for Disease Control and
Prevention Compressed Mortality Files (1999–2002)
(Center for Disease Control and Prevention). The
correlation with state income inequality was
r ¼ 0.61 (p o0.01). As for education and imprisonment (above), DC was an outlier, but the association remained statistically significant when it was
excluded. It appears that there are no reliable
international data on drug-related deaths (Advisory
Council on the Misuse of Drugs, 2000).
Social mobility
International data on intergenerational social
mobility are available for a few countries from a
study by Blanden, Gregg, and Machin (2005). Social
mobility was measured by estimating the correlation
between father’s and son’s incomes (when sons were
close to age 30) and calculated from large, representative cohort studies in each of eight countries.
Higher correlations between father’s and son’s
incomes therefore indicate less social mobility.
Despite having data for only eight countries, the
relationship between intergenerational social mobility and income inequality was statistically significant (r ¼ 0.93, po0.01). The relationship is shown
in Fig. 5: among these eight countries bigger income
differences are associated with lower social mobility.
When the USA and UK are excluded as possible
outliers, the correlation remains close (r ¼ 0.60) but
with only six data points is not statistically
significant.
Discussion
The evidence outlined here, consistent as it is
across outcomes and setting, goes some way to
establishing the simple but important point that
numerous social problems associated with relative
deprivation—from ill health to poorer educational
performance—are more common in more unequal
societies.
For comparability with earlier work the new
analyses presented in this paper used Pearson
correlation coefficients, which assume normal distributions and linearity. However, we found that the
use of non-parametric Spearman rank correlations
produced broadly similar results and in no instance
affected statistical inference. We also checked to see
if the international results were robust to the use not
only of the ratio of the top to bottom 20% of
incomes, but also to the ratio of the top and bottom
10% and to the Gini coefficient. In all cases the
measure of inequality made no substantive difference to the results.
Whilst causal inference from observational studies, particularly from ecological studies, is inherently problematic, this body of evidence meets
epidemiological guidelines for assessing causality
(Gordis, 2004; Hill, 1965). Previous studies of
Fig. 5. Intergenerational social mobility in relation to income inequality among rich countries.
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
associations between income inequality and health
have shown that these are strong relationships and
that they exhibit a dose-response form; as inequality
increases so does poor health (Wilkinson & Pickett,
2006). Findings at the scale of nations and states
(but not more local areas) have been well replicated
(Subramanian & Kawachi, 2004; Wilkinson &
Pickett, 2006), and the biology of chronic stress
provides a plausible biological explanation of the
findings (Sapolsky, 2005). As well as several time
series analyses (Marmot & Bobak, 2000; Wilkinson,
1992), there are also examples of societies—such as
Britain during the two World Wars and the
formerly centrally planned societies undergoing
transition to market economies (Wilkinson,
1996)—showing that changes in income inequality
are followed by changes in health outcomes. What
our new evidence adds is coherence and specificity:
if relative deprivation links income inequality to
poor health then we would expect to find, as we do,
that other social problems linked to relative
deprivation are also associated with income inequality but are not associated with absolute levels of
income as such.
This picture suggests that more unequal societies
are socially dysfunctional in many different ways. It
is striking that a group of more egalitarian countries
(usually including Japan, Sweden, Norway and
often other Scandinavian countries) perform well
on a variety of outcomes, and a similar group of
more unequal countries (including the USA, Portugal and often the UK) tend to have poorer
outcomes. We find it hard to think of other possible
explanations—apart from inequality—for these
patterns. Although there are clearly numerous
similarities between the US and Britain, the same
cannot be said of the US and Portugal or Singapore
which also tend to perform badly. Singapore and
the US are ethnically diverse, but Portugal has a
more homogeneous population like Finland or
Japan. Japan and the US are service-based economies, while Singapore and Finland have a heavier
manufacturing base. Despite performing well on
almost all outcomes, Sweden and Japan show
marked contrasts in the position of women in
society and in their participation in paid employment. In addition, Japan—in contrast to Sweden—
has low rates of single parenthood and divorce.
Even their pathways to greater equality of incomes
differ substantially: Sweden depends primarily on
redistribution through taxes and benefits while
Japan has smaller differences in earnings even
1973
before taxes and benefits. Sweden has one of the
strongest welfare systems, while Japan devotes an
unusually small proportion of its National Income
to public social expenditure. Not even unemployment rates fit the pattern: Finland has an unemployment rate of 8.4%, much closer to Portugal’s
7.6% than to Japan’s 4.4%, which is more like
Singapore’s 3.1% (Central Intelligence Agency,
2006).
Interpretation
Human beings have lived in every kind of society,
ranging from the most egalitarian foraging societies
of prehistory to the most tyrannical hierarchies
(Boehm, 1993; Erdal & Whiten, 1996). Modern
societies almost certainly continue to differ in how
hierarchical and socially differentiated they are. The
development of comparable measures of income
inequality seems at last to have given us a rough
indication of the extent of social differentiation
from one society to another, and so a new approach
to a vital arena for social research.
Despite having data for only eight countries, the
apparent tendency for societies with wider income
differences to have less social mobility, seems to
confirm the relation between income inequality and
social stratification. Bigger income differences seem
to solidify the social structure and decrease the
chances of social mobility. In effect, equal opportunity is a more distant prospect where there are
greater inequalities of outcome. Reinforcing this
impression is the fact that while income differences
widened in Britain and the United States, social
mobility slowed (Blanden et al., 2005) and, as if
greater social distances were translated into greater
geographical distances, residential segregation of
rich and poor increased (Berube, 2005; Kawachi,
2002; Mayer, 2001).
As if to confirm that the link between income
inequality and these outcomes is indeed mediated by
changes in the burden of relative deprivation, there
are indications that the outcomes most strongly
related to deprivation and showing steeper social
gradients within societies, are also those most
closely related to income inequality. For example,
the relation between population mortality rates and
income distribution is typically strongest among
men of working age; this is also the age and sex
group in which the social class gradient in health
tends to be steepest. Similarly, just as income
inequality is more closely related to the population
ARTICLE IN PRESS
1974
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
prevalence of obesity among women than men, so
too is the social gradient in obesity more marked
among women than men (Molarius, Seidell, Sans,
Tuomilehto, & Kuulasmaa, 2000). Violence and
teenage births also seem to be examples of problems
which are particularly strongly related to both
relative deprivation and to income inequality
(Health Development Agency, 2003; Wilson &
Daly, 1997). These observations seem to confirm
the simplest interpretation: that the reason why
greater inequality is associated with a higher
prevalence of these problems is that they are partly
responses to the burden of relative deprivation, and
inequality increases that burden.
The relation between inequality and a range of
health and social problems may also contribute to
the debate around theories of social selection versus
social causation in the production of social gradients in these problems (see for example: Claussen,
Smits, Naess, & Davey Smith, 2005; Dohrenwend
et al., 1992; Goldman, 1994; Hudson, 2005; McMunn,
Bartley, Hardy, & Kuh, 2006; Ritsher, Warner,
Johnson, & Dohrenwend, 2001). According to
social selection theory, social gradients may reflect
a tendency for social mobility to discriminate
between the healthy and the unhealthy, so that the
healthiest and most capable people move up the
social ladder and end up with the highest incomes,
while the least healthy end up at the bottom of the
income distribution. Alternatively, social gradients
in health might result from the way people’s health
risks are shaped by less favourable social and
material circumstances. If we can demonstrate a
strong relationship between national levels of
income inequality and a range of social problems,
beyond health, we might gain a new perspective on
the problem of social selection versus social causation, in relation to health. If we observe more
homicides or a higher prevalence of obesity, in more
unequal societies, it is unlikely that having a higher
level of violence or a greater proportion of heavy
people has caused income distributions to be wider.
It is instead much more likely that causation runs
the other way—wider income differences leading to
more violence and more obesity. And if greater
economic inequality in a society is associated with
more of most of the problems associated with
relative deprivation, then this strongly suggests the
operation of powerful processes of social causation
for health as well. This does not mean that social
mobility does not also act selectively: it may in turn
select according to how people have been shaped by
greater inequality. This is particularly plausible in
relation to the long-term effects of circumstances in
early life.
As well as sharing roots in relative deprivation,
many social problems—including poor health—may
also involve similar causal processes involving
psychosocial pathways related to chronic stress.
The relative contributions to the link between
inequality and population health of psychosocially
mediated, or of the direct, unmediated, effects of
material factors, have been disputed (Lynch, Smith,
Kaplan, & House, 2000; Marmot & Wilkinson,
2001). However, many of the social problems, which
seem to be related to income inequality are
inherently behavioural and provide evidence that
income inequality has psychosocial effects. Indeed,
if low social status increases chronic stress, this
would provide fertile ground suggesting why so
many health and behavioural problems seem rooted
in relative deprivation and show similar social
gradients and relationships with income inequality.
Researchers have frequently assumed that if
income differences are important, this implies the
importance of processes of social comparison. A
long tradition of sociological theory, stretching
back to Marx’s assumptions about the importance
of the larger social framework (Bergesen & Bata,
2002) and Durkheim’s notion of ‘‘social facts’’
(Berkman, Glass, Brissette, & Seeman, 2000;
Schwartz & Diez-Roux, 2001), has emphasised the
ways in which individuals are shaped by the social
environment, yet empirical investigation of the
scales and pathways that underpin social comparisons is lacking. Following Runciman (1966), such
comparisons are often thought to take place
primarily at the local level, although there is some
recognition that those ‘‘who are spatial neighbours
are not always social neighbours’’ (Mitchell, Gleave,
Bartley, Wiggins, & Joshi, 2000). But if income
inequality is related to health more closely when the
units of analysis are whole societies than when they
are smaller areas, this implies that we are dealing
with effects of relativities across a much larger scale.
Dunn, Veenstra, and Ross (2006) recently reported
relationships between health and various measures
of socioeconomic status in a Canadian survey. They
found that self-rated health was predicted by
whether or not people felt they were better or worse
off than the average Canadian. They also calculated
how people’s actual incomes compared in relation
to incomes in their neighbourhood and in their
province. Although statistically significant for both,
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
the correlations with health were stronger when
people’s incomes were related to province-level
income than they were when related to neighbourhood income. The less local provincial reference
point was more salient than the neighbourhood
reference point. This fits with what we would expect,
based on our review of studies of income inequality
and health (Wilkinson & Pickett, 2006).
Because of the tendency to expect reference points
for income to be local, people frequently refer to
actual levels of monetary income as ‘‘absolute’’
income. But within a single nation it is of course
impossible to tell whether relationships between
health and income arise because of the absolute
material standards of living which any given level of
income buys, or because of where it places you in
the national status hierarchy. Only when we
compare income in different societies, as in our
comparison of black Americans and Costa Rican’s
above, or when we look at life expectancy and
Gross National Income per capita among the
richest countries, do we see how weak the relation
between absolute income levels and population
health may be. Yet it is still not uncommon to refer
to associations between the mean income and health
for small areas as indicating effects of ‘‘absolute’’
income, rather than the effects of how the local
population fits into the scale of national relativities—whether the area is deprived or not in relation
to the standards of the rest of society.
What we are suggesting may look like action at a
distance. Relativities and comparisons beyond the
local seem more important than purely local ones.
This looks less implausible if we are thinking of
income distribution as a determinant of processes of
social class differentiation rather than simply in
terms of income comparisons. As we said above, the
range of health and social problems, which seem to
be related to income distribution, and the fact that
they are also related to relative deprivation, suggests
that the salience of income distribution involves
something much deeper than comparisons of
income. When income distribution first appeared
on the public health agenda, there was a tendency to
assume that, if it was related to health at all, it must
reflect some additional and previously unknown
causal process. But it now seems much more likely
that it is telling us more about the nature of the
factors driving the familiar social class differences in
health. What matters is perhaps that the associations between inequality and the prevalence of
different outcomes are telling us more about the
1975
causal processes rooted in social differentiation. If,
in addition, income distribution is really telling us
about more general processes of social class
differentiation, then it may be mistaken to analyse
its effects after adjusting for other factors closely
related to class, such as education.
An indication of the range of the processes that
may be set in motion by greater social differentiation comes from the data shown in Figs. 3 and 4 on
imprisonment. Analyses of the increase in prison
populations in Britain and the United States during
the later 20th century, when income differences were
widening, suggest that the greater part of the
increase resulted from more punitive sentencing
rather than from increases in crime (Mauer, 2001).
A relation between wider income differences and
more punitive attitudes to crime is suggested in
Fig. 4 which shows, cross-sectionally, that more
unequal states are more likely to retain the death
penalty. Greater inequality and bigger social distances may then be accompanied by hardening of
social attitudes.
The remaining question is of course how one
knows one’s class or social status in the wider
society. The answer is likely to involve a knowledge
of where one fits into many different relativities, all
substantially influenced by income. A person may
assess these relativities in relation to those she was
at school with, where her school fitted into the social
hierarchy, the income and social standing of her
parents, the social connotations of her house and
the part of town in which she lives, her educational
achievements, place in the job hierarchy, and her
knowledge—gained through the media—of the lives
of the elite, the rich and powerful, the celebrities,
and so on. Somehow, we all learn the degrees of
superiority and inferiority in our society and know
where we stand. As Emerson (1883) said, ‘‘Tis very
certain that each man carries in his eye the exact
indication of his rank in the immense scale of men,
and we are always learning to read it’’.
Conclusions
It is often assumed that the desire to raise
national standards of performance in fields such as
education and health is a quite separate problem
from the desire to reduce health and educational
inequalities within a society. However, perhaps the
most important implication of the relationships
with inequality shown here is that the achievement
of higher national standards of performance may be
ARTICLE IN PRESS
1976
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
substantially dependent on reducing inequalities in
each country. As well as improving health, reducing
inequality may also raise the educational performance of school children, increase trust, while
decreasing violence and teenage births.
The associations we have seen between income
inequality and a range of health and social problems
are far from trivial. Even ignoring extreme examples, there are ten-fold differences in homicide rates
between more and less equal countries and US
states, six-fold differences in teenage birth rates, sixfold differences in the prevalence of obesity, fourfold differences in how much people feel they can
trust each other, five- or ten-fold differences in
imprisonment rates and, mainly as a result of deaths
at younger ages, 3 years difference in the average
length of life. These issues go to the heart of
problems which beset our societies and are constantly in the news. They attract strings of policy
initiatives designed to tackle each of these issues
separately: policies to reduce overcrowded prisons,
to reduce violence or teenage births, to raise
children’s educational performance and so on as if
there was no connection between them.
Although Britain is said to be ‘‘ahead of
continental Europe in developing and implementing
policies to reduce socio-economic inequalities in
health’’ (Mackenbach, 2006b), so far they have met
with little success (Department of Health, 2005;
Mackenbach, 2006a). Perhaps that is because they
have focussed less on decreasing the burden of
relative deprivation than on attempts to reduce its
effects on health. It is difficult to stop relative
deprivation having its familiar effects on health.
However, if ill health is just one of the many social
problems related to relative deprivation, which is
less common in more egalitarian countries, then
there are likely to be substantial and widespread
benefits of tackling the underlying inequality itself.
Indeed, if inequality has psychosocial effects,
perhaps involving chronic stress and the aversive
effects of low social status, then it is possible that
some of the health and social problems marked by
social gradients share roots in chronic stress. Rather
than providing ever more prisons, doctors, health
promoters, social workers, educational psychologists, and drug rehabilitation units, in expensive and
at best only partially effective attempts to offset the
problems of relative deprivation, it may be cheaper
and more rewarding to tackle the underlying
inequalities themselves. The differences in inequality
we have been looking at are, after all, not
differences between an impractical perfect equality
and practical reality. Instead, they seem to show the
importance of the existing differences in inequality
which occur now between developed market
democracies or between US states, and which can
only be revealed through comparative analysis at
the scale of whole societies.
References
Advisory Council on the Misuse of Drugs. (2000). Reducing drugrelated deaths. London: Stationary Office.
Allison, P. D. (1978). Measures of inequality. American Sociological Review, 43, 865–880.
Ballas, D., Dorling, D., & Shaw, M. (2007). Social inequality,
health, and well-being. In J. Hawrorth, & G. Hart (Eds.),
Well-being: Individual, community, and social perspectives.
Basingstoke: Palgrave.
Ben-Shlomo, Y., & Kuh, D. (2002). A life course approach to
chronic disease epidemiology: Conceptual models, empirical
challenges and interdisciplinary perspectives. International
Journal of Epidemiology, 31(2), 285–293.
Bergesen, A. J., & Bata, M. (2002). Global and national
inequality: Are they connected? Journal of World-Systems
Research, III(I), 130–144.
Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000).
From social integration to health: Durkheim in the new
millennium. Social Science & Medicine, 51(6), 843–857.
Berube, A. (2005). Mixed communities in England. York: Joseph
Rowntree Foundation.
Blanden, J., Gregg, P., & Machin, S. (2005). Intergenerational
mobility in Europe and North America. London: Centre for
Economic Performance, London School of Economics.
Boehm, C. (1993). Egalitarian behavior and reverse dominance
hierarchy. Current Anthropology, 34, 227–254.
Bourdieu, P. (1984). Distinction. Cambridge: Harvard University
Press.
Brenner, N. (2001). The limits to scale? Methodological reflections of scalar structuration. Progess in Human Geography,
25(4), 591–614.
Center for Disease Control and Prevention Compressed Mortality Files (1999–2002). Accessed on: 30 March 2006 Available
at: /http://wonder.cdc.gov/wonder/help/mort.htmlS.
Central Intelligence Agency World Factbook Accessed on: 13
September 2006. Available at: /http://www.cia.gov/cia/
publications/factbook/index.htmlS.
Claussen, B., Smits, J., Naess, O., & Davey Smith, G. (2005).
Intragenerational mobility and mortality in Oslo: Social
selection versus social causation. Social Science & Medicine,
61(12), 2513–2520.
de Tocqueville, A. (2000). Democracy in America. Indianapolis:
Hackett.
Demyttenaere, K., Bruffaerts, R., Posada-Villa, J., Gasquet, I.,
Kovess, V., Lepine, J. P., et al. (2004). Prevalence, severity,
and unmet need for treatment of mental disorders in the
World Health Organization World Mental Health Surveys.
Journal of American Medical Association, 291(21), 2581–2590.
Department of Health. (2005). Tackling health inequalities: Status
report on the programme for action. London: Stationary
Office.
ARTICLE IN PRESS
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
Diez-Roux, A. (1998). Bringing context back into epidemiology:
Variables and fallacies in multi-level analysis. American
Journal of Public Health, 88, 216–222.
Dohrenwend, B. P., Levav, I., Shrout, P. E., Schwartz, S.,
Naveh, G., Link, B. G., et al. (1992). Socioeconomic status
and psychiatric disorders: The causation-selection issue.
Science, 255(5047), 946–952.
Dunn, J. R., Veenstra, G., & Ross, N. (2006). Psychosocial and
neo-material dimensions of SES and health revisited: Predictors of self-rated health in a Canadian national survey.
Social Science & Medicine, 62(6), 1465–1473.
Emerson, R. W. (1883). The conduct of life (Originally published
by Macmillan, London), 1st World Library 2005 edition.
Erdal, D., & Whiten, A. (1996). Egalitarianism and Machiavellian intelligence in human evolution. In P. Mellars, & K.
Gibson (Eds.), Modelling the early human mind. Cambridge:
McDonald Institute Monographs.
Franzini, L., Ribble, J., & Spears, W. (2001). The effects of
income inequality and income level on mortality vary by
population size in Texas counties. Journal of Health and
Social Behavior, 42(4), 373–387.
Gold, R., Kawachi, I., Kennedy, B. P., Lynch, J. W., &
Connell, F. A. (2001). Ecological analysis of teen birth rates:
Association with community income and income inequality.
Maternal and Child Health Journal, 5(3), 161–167.
Gold, R., Kennedy, B., Connell, F., & Kawachi, I. (2002). Teen
births, income inequality, and social capital: Developing an
understanding of the causal pathway. Health and Place, 8(2),
77–83.
Goldman, N. (1994). Social factors and health: The causationselection issue revisited. Proceedings of the National Academy
of Science USA, 91(4), 1251–1255.
Gordis, L. (2004). Epidemiology. Philadelphia: Elsevier Saunders.
Health Development Agency. (2003). Teenage pregnancy
and parenthood: A review of reviews. Evidence briefing.
London: Health Development Agency, National Health
Service.
Hill, A. B. (1965). The environment and disease: Association or
causation? Proceedings of the Royal Society of Medicine, 58,
295–300.
Hsieh, C.-C., & Pugh, M. D. (1993). Poverty, income inequality,
and violent crime: A meta-analysis of recent aggregate data
studies. Criminal Justice Review, 18, 182–202.
Hudson, C. G. (2005). Socioeconomic status and mental illness:
Tests of the social causation and selection hypotheses.
American Journal of Orthopsychiatry, 75(1), 3–18.
Kawachi, I. (2002). Income inequality and economic residential
segregation. Journal of Epidemiology and Community Health,
56(3), 165–166.
Kawachi, I., Kennedy, B. P., Lochner, K., & Prothrow-Stith, D.
(1997). Social capital, income inequality, and mortality.
American Journal of Public Health, 87(9), 1491–1498.
Kennedy, B. P., Kawachi, I., Lochner, K., Jones, C., &
Prothrow-Stith, D. (1997). (Dis)respect and black mortality.
Ethnicity and Disease, 7(3), 207–214.
Lobmayer, P., & Wilkinson, R. G. (2002). Inequality, residential
segregation by income, and mortality in US cities. Journal of
Epidemiology and Community Health, 56(3), 183–187.
Lynch, J. W., Smith, G. D., Kaplan, G. A., & House, J. S. (2000).
Income inequality and mortality: Importance to health of
individual income, psychosocial environment, or material
conditions. British Medical Journal, 320(7243), 1200–1204.
1977
Macintyre, S., Maciver, S., & Sooman, A. (1993). Area, class and
health: Should we be focusing on places or people? Journal of
Social Policy, 22(2), 213–234.
Mackenbach, J. P. (2006a). Health Inequalities: Europe in Profile.
Report for ‘Tackling Health Inequalities Governing for Health.
London: Central Office of Information.
Mackenbach, J. P. (2006b). Socio-economic inequalities in health
in Western Europe. In J. Siegrist, & M. Marmot (Eds.), Social
inequalities in health. Oxford: Oxford University Press.
Marmot, M., & Bobak, M. (2000). International comparators
and poverty and health in Europe. British Medical Journal,
321(7269), 1124–1128.
Marmot, M., & Wilkinson, R. G. (2001). Psychosocial and
material pathways in the relation between income and health:
A response to Lynch et al. British Medical Journal, 322(7296),
1233–1236.
Marston, S. A. (2000). The social construction of scale. Progress
in Human Geography, 24(2), 219–242.
Marston, S. A., Jones, J. P., & Woodward, S. A. (2005). Human
geography without scale. Transactions of Institute of British
Geographers, 30, 416–432.
Mauer, M. (2001). The causes and consequences of prison growth
in the United States. Punishment & Society, 3(1), 9–20.
Mayer, S. (2001). How the growth in income inequality increased
economic segregation. The Joint Center for Poverty Research
Working Paper 235. Chicago: NorthWestern University/
University of Chicago.
McMunn, A., Bartley, M., Hardy, R., & Kuh, D. (2006). Life
course social roles and women’s health in mid-life: Causation
or selection? Journal of Epidemiology and Community Health,
60(6), 484–489.
Merlo, J. (2003). Multilevel analytical approaches in social
epidemiology: Measures of health variation compared with
traditional measures of association. Journal of Epidemiology
and Community Health, 57(8), 550–552.
Mitchell, R., Gleave, S., Bartley, M., Wiggins, D., & Joshi, H.
(2000). Do attitude and area influence health? A multilevel
approach to health inequalities. Health and Place, 6(2), 67–79.
Molarius, A., Seidell, J. C., Sans, S., Tuomilehto, J., &
Kuulasmaa, K. (2000). Educational level, relative body weight
and changes in their association over 10 years: An international perspective from the WHO MONICA project. American Journal of Public Health, 90, 1260–1286.
OECD Programme for International Student Assessment. (2004).
Learning for tomorrow’s world? First results from PISA, 2003.
Pickett, K. E., James, O. W., & Wilkinson, R. G. (2006). Income
inequality and the prevalence of mental illness: A preliminary
international analysis. Journal of Epidemiology and Community Health, 60(7), 646–647.
Pickett, K. E., Kelly, S., Brunner, E., Lobstein, T., &
Wilkinson, R. G. (2005). Wider income gaps, wider waistbands? An ecological study of obesity and income inequality.
Journal of Epidemiology and Community Health, 59(8),
670–674.
Pickett, K. E., Mookherjee, J., & Wilkinson, R. G. (2005).
Adolescent birth rates, total homicides, and income inequality
in rich countries. American Journal of Public Health, 95(7),
1181–1183.
Pickett, K. E., & Pearl, M. (2001). Multilevel analyses of
neighbourhood socioeconomic context and health outcomes:
A critical review. Journal of Epidemiology and Community
Health, 55(2), 111–122.
ARTICLE IN PRESS
1978
R.G. Wilkinson, K.E. Pickett / Social Science & Medicine 65 (2007) 1965–1978
Putnam, R. D. (1993). Making democracy work: Civic traditions
in modern Italy. Princeton: Princeton University Press 224pp.
Putnam, R. D. (2000). Bowling alone: The collapse and revival of
American community. New York: Simon and Schuster.
Ritsher, J. E., Warner, V., Johnson, J. G., & Dohrenwend, B. P.
(2001). Inter-generational longitudinal study of social class
and depression: A test of social causation and social selection
models. British Journal of Psychiatry Supplementary, 40,
s84–s90.
Runciman, W. G. (1966). Relative deprivation and social justice: a
study of attitudes to social inequality in twentieth-century
England. London: Routledge.
Sapolsky, R. (2005). Sick of poverty. Scientific American, 293(6),
92–99.
Schwartz, S., & Diez-Roux, A. V. (2001). Commentary: Causes
of incidence and causes of cases—A Durkheimian perspective
on Rose. International Journal of Epidemiology, 30(3),
435–439.
Subramanian, S. V., & Kawachi, I. (2004). Income inequality and
health: What have we learned so far? Epidemiologic Reviews,
26, 78–91.
Taylor, P. J., & Flint, C. (2000). Political geography: Worldeconomy, nation-state and locality. Harlow: Prentice-Hall.
Tunstall, H. V., Shaw, M., & Dorling, D. (2004). Places and health.
Journal of Epidemiology and Community Health, 58(1), 6–10.
United Nations Crime and Justice Information Network. (2000).
Survey on crime trends and the operations of criminal justice
systems (fifth, sixth, seventh): United Nations.
United Nations Development Program. (2003). Human development report. New York: Oxford University Press.
US Census Bureau. (2004). Table S4. Gini ratios by State: 1969,
1979, 1989, 1999: Income Statistics Branch/HHES Division.
US Department of Education, N.C.f.E.S. (2004a). The Nation’s
report card: Mathematics highlights 2003. Washington, DC.
US Department of Education, N.C.f.E.S. (2004b). The Nation’s
report card: Reading highlights 2003. Washington, DC.
US Department of Justice, B.o.J.S. Incarceration rates for
prisoners under State or Federal jurisdiction. File: corpop25.wk1. Accessed on 30 March 2006 Available at:
/http://www.ojp.usdoj.gov/bjs/data/corpop25.wk1S.
Uslaner, E. (2002). The moral foundations of trust. Cambridge:
Cambridge University Press.
Wilkinson, R. G. (1992). Income distribution and life expectancy.
British Medical Journal, 304(6820), 165–168.
Wilkinson, R. G. (1996). Unhealthy societies: The afflictions of
inequality. London: Routledge.
Wilkinson, R. G. (1997). Comment: Income, inequality, and
social cohesion. American Journal of Public Health, 87(9),
1504–1506.
Wilkinson, R. G. (2005). The impact of inequality. New York:
New Press.
Wilkinson, R. G., & Pickett, K. E. (2006). Income inequality and
population health: A review and explanation of the evidence.
Social Science & Medicine, 62(7), 1768–1784.
Williams, R. B., Feaganes, J., & Barefoot, J. C. (1995). Hostility
and death rates in 10 US cities. Psychosomatic Medicine,
57(1), 94.
Williams, S. J. (1995). Theorizing class, health and lifestyles: Can
Bourdieu help us? Sociology of Health and Illness, 17,
577–604.
Wilson, M., & Daly, M. (1997). Life expectancy, economic inequality, homicide, and reproductive timing in
Chicago neighbourhoods. British Medical Journal, 314(7089),
1271–1274.