Content uploaded by Stefan Liebig
Author content
All content in this area was uploaded by Stefan Liebig on Jul 13, 2015
Content may be subject to copyright.
1
Richard Wilkinson and Kate Pickett (2009): The Spirit Level. Why More
Equal Societies Almost Always Do Better. Allen Lane.
Reviewed by Stefan Liebig*
The Spirit Level. Why More Equal Societies Almost Always Do Better by the British
epidemiologists Richard Wilkinson and Kate Pickett found broad resonance in the media
upon its publication in 2009. This is not surprising, since it makes a very simple and, at
first sight, alluring statement: by reducing income differences in a society, it is possible to
eliminate all social problems and ills. Income equality reduces people’s stress levels and
prevents the associated physical and mental illnesses; members of society trust and
support one another, live in harmonious relationships, do not take drugs, are not
overweight, are more intelligent, and do not have to commit criminal acts. Who would
not want to live in a world like this? Yet whenever simple solutions are offered for
complex problems and correlations, a degree of skepticism is healthy. And that is true of
this book in particular.
The basic idea is to use empirical data to answer the normative question of how
goods should be distributed in a society. This sounds like a case of the naturalistic
fallacy—that is, deriving a normative statement from a descriptive one—but on closer
inspection, it is not. The call for equal distribution of social goods presented in the book
is derived from the notion that inequality reduces the wellbeing of all (sic!) members of a
society. This means that decreasing inequality appears to be a means of increasing the
general wellbeing and welfare of a society. Equality or inequality of the distribution of
wealth is therefore not a matter of morality. If there is sufficient guarantee that inequality
has negative consequences for individual or social wellbeing, political acumen requires
us to strive for an egalitarian society. This is precisely what the authors aim to
demonstrate.
The method to be used is clear: take several societies—in this particular case, the
authors concentrate on the 23 richest countries—that differ in their degree of social
inequality, then examine whether a correlation between the level of inequality and a
number of different indicators of individual wellbeing and social welfare may be
observed. The first thing needed is a way of measuring social inequality. Although
understanding social inequality as a multidimensional construct is now an integral part of
sociological research on social inequality (the relevant dimensions being, for instance,
income, power, prestige, education, or social security) and a distinction must be drawn
between inequalities in distribution and inequalities of opportunity, Wilkinson and
Pickett confine themselves to income. They justify this decision by arguing that there are
no other indicators available for international comparative analyses—which does initially
seem to make sense. To measure the extent of income inequality, they choose the ratio of
income of the top 20 percent to the lower 20 percent in a society, drawing on data from
the United Nations Human Development Report. The 23 countries are ranked according
to their degree of income inequality, the Scandinavian countries and Japan representing
* Collaborative Research Center 882: From Heterogeneities to Inequalities, Department of
Sociology, Bielefeld University, Germany.
2
one extreme with the lowest income inequality, and the UK, Portugal, USA, and
Singapore the other extreme with the highest income inequality.
On the basis of this ranking, the authors then present a correlation between the
degree of income inequality and a total of nine social problems, which are identified as
“costs of inequality”: (1) community life and social relations, (2) mental health and drug
use, (3) physical health and life expectancy, (4) obesity, (5) educational performance, (6)
teenage pregnancies, (7) violence, (8) crime and punishment, and (9) unequal
opportunities for intergenerational social mobility. The authors refer to the findings of a
large number of epidemiological studies, social science surveys, and OECD and UN data.
However, the crux of their argument is always presented in a neat graph, in which they
show the degree of income inequality on the x-axis and the extent of the relevant social
problems—normally the mean values for the various countries—on the y-axis. The result
is always a “point cloud” formed by the individual countries, through which the authors
draw a straight line. The gradient of this line is intended to demonstrate the correlation
between income inequality and the relevant social problem.
For each of the nine social problems—just as to be expected after reading the
introductory chapters—a correlation with the degree of a society’s income inequality
emerges: countries with lower income inequality show a low level of the various
problems, while in countries with higher income inequality people have less trust and a
higher level of mental illnesses, consume more drugs, have worse general health and
lower life expectancy, the rate of obesity is higher and educational performance is lower,
there are more teenage pregnancies, there is more violence and crime, prisons are
overcrowded, and children from disadvantaged sections of the population have poorer
career prospects.
These findings, at first glance impressively consistent, immediately raise the
question whether the extent of income inequality alone—in the sense of a
monocausality—can in fact be held accountable for all this. Wilkinson and Pickett feel
compelled to draw that conclusion. They justify it by linking together two arguments.
The first is that the social problems observed at the level of society as a whole can be
attributed to increased status competition and social anxiety among individual members
of society. People who constantly find themselves competing with others not only
develop psychological symptoms of stress, which then turn into physical illnesses in the
long term, but are also more suspicious of others, more likely to resort to sedative or
stimulant drugs and—driven by fears of social exclusion and the experience of social
contempt—tend towards violence and delinquent behavior, etc. The second argument is
that a high degree of income inequality describes a societal situation in which there is a
high level of status competition. As inequality in a society increases, there is more
competition between individual members of the society, the risk of status loss is greater,
and social anxieties are exacerbated. All in all, this means that income inequality
increases status competition and social anxiety, which in turn triggers stress responses in
individuals and is expressed in an increase of the above-mentioned social problems at the
level of society as a whole. Conversely, it means if we want a society without these social
problems, we will have to establish an egalitarian distribution of income, because this
will enable us to reduce the status competition and social anxieties responsible for the ills
identified. Hence, an egalitarian society is demanded not by morality, but—according to
the empirical findings—by political acumen.
3
This is further corroborated by a third argument: for Wilkinson and Pickett, an
egalitarian distribution of income is essential also because it not only helps those who are
less well off but produces welfare effects for all citizens. The daily life of those who are
materially better off is also threatened by a high degree of competition and status anxiety.
They too develop symptoms of physical and psychological illness, as is evident from the
generally poor state of health of the entire population of societies with high levels of
income inequality. And, of course, it is also in the interest of this part of the population if
a community does not have to allocate as much public funding to combating social
problems (crime, drug use, high morbidity, etc.) and can reduce taxes instead.
At first, this argumentation closely follows a explanatory model that is very
common in social sciences. In a first step, the question is asked how structural conditions
in a society—e.g. the distribution of income—at the macro level determine the everyday
situation, the opportunities and restrictions of its individual members on the micro level.
In a second step, it is shown how individuals behave in view of these social conditions
(micro level), in order then, in a third step, to identify the social consequences on the
macro level that result when all or some of the members of society display a particular
type of behavior. Using this macro-micro-macro-model of explanation it is in fact
possible to discover causal links between social phenomena, because it is possible to
specify in more detail the mechanisms by which social phenomena emerge and/or are
changed by individual actions and behavior. In this respect, no objections can be made to
Wilkinson and Pickett’s argumentation strategy. But there is a substantial gap between
theoretical reasoning and empirical proof: As the authors are exclusively using
aggregated data on the societal or macro level they are not able to test their “micro”
assumptions, how the distribution of income in a society really affects individual
behavior – for this endeavor one needs comparative data on the micro level i.e. survey
data. What is more important, however, is that in their second argument they make a
link which does not hold as such. They equate income inequalities with status inequalities,
inferring a high level of status inequality from the existence of high income inequality.
But this is precisely the equation that cannot be made—because in modern societies a
high income is not synonymous with high status, and vice versa. Inferring high status
competition in a society from a high level of income inequality in that society is also
inadmissible, as there are societies with high status inequality and thus high status
competition despite lower income inequality. The classic example is Japan. Consequently,
Wilkinson and Pickett’s explanation is not really convincing. It then becomes
questionable whether the empirical findings presented can in fact be attributed
consistently and solely to the phenomenon of income inequality—particularly since so
many examples of “spurious correlations” can be found in the natural and social sciences
(such as the famous example of the stork and the newborn baby).
Do the consistent empirical findings not speak for themselves? Are they not
enough to demonstrate that egalitarian societies are better societies? This is precisely
where a methodological critique is called for. To begin with a fundamental point, the
findings presented all show only correlations between two variables. This means firstly
that they cannot be interpreted in terms of a direct causal link. Secondly, the correlations
are established only very imprecisely in the graphs: to show the strength of a correlation,
it is not sufficient to draw a somewhat associatively selected straight line through a point
4
cloud. Sophisticated statistical methods are now available that would need, at least, to be
cited as evidence of the findings depicted in the graphs. Furthermore, correlations are not
necessarily linear and constant: they can perfectly well—and this applies to many social
phenomena—take quite different courses. We also know that statistical correlations are
strongly determined by individual extreme cases or “outliers.” What is particularly
striking about the findings reported is that, for instance, the US represents the most
negative pole for most social problems, but other countries show a considerably higher
level of income inequality. At the same time, there are a number of examples in which
countries differing dramatically in income inequality show either the same level of the
social problems studied or else a correlation contrary to that set out by the authors.
Secondly, to prove the stated causal effects, it is not sufficient to show this on the
basis of bivariate, cross-sectional findings: the correlation between two variables may
also be caused by a third variable which exerts its effect indirectly via one of the others.
This is very plausible for at least some of the correlations reported in the book. Hence,
the differences in the general state of health and in educational participation or
performance might also be explained by the very different institutional structure of the
national health and education systems. Cultural differences between the countries, too,
may be responsible for the differences observed. This applies, for example, to eating
habits and the findings on obesity. As opposed to the linear correlation Wilkinson and
Pickett believe they see, it is quite possible to identify a pattern of “clustering” according
to cultural similarity: central and northern European vs. southern European. Controlling
for third variables of this kind, and thus allowing the influence of income inequality on
the relevant social problems to be unambiguously identified, requires more complex
statistical methods. This book therefore suggests a lack of ambiguity in the findings that
is not in fact the case.
At the same time, it is now a generally held view that causal effects in social
sciences can only be clearly identified using longitudinal data. It must thus be shown that
a change in income inequality in one country has the postulated consequence for the
relevant social problems. Since Wilkinson and Pickett claim that a reduction of income
inequality would alleviate the social problems in a society, this is the effect they need to
demonstrate. Since income inequality in the 23 countries studied has changed over time,
particularly since the 1980s (see OECD 2008), this would be feasible in principle. A
reduction in income inequality would then have to have led to—delayed—effects
pertaining to the social problems. This type of longitudinal findings are not reported,
however.
A number of other methodological difficulties could be cited—for instance, using
only needs-adjusted net household income is problematic as the resulting inequality
measure is not only a result of the income distribution but also of the distribution of
household sizes which differs between societies.
Overall, this book thus does not provide adequate—in the sense of
methodologically sound—empirical findings which, firstly, allow an interpretation of the
observed social problems that attributes them solely to income inequalities and, secondly,
support the assumption that a reduction of income inequality would lead to a reduction of
the social problems identified.
The basic claim of this book is, that egalitarian societies are just societies. Of
course, it is possible to do this from a particular normative position, but the debate in
5
political philosophy and the results from empirical research show that justice does not
necessarily mean equality. Nor can people’s understanding of justice be reduced to
equality. On the contrary, a central pillar of the concept of justice in modern societies is
the idea of rewarding individual performance. Just as there are differences in individual
performance and effort, so the rewards must also be different. And, in fact, in national or
even international comparative surveys only a tiny minority of respondents advocate
paying wages and salaries irrespective of people’s individual effort and performance. The
same pay for different levels of performance is consequently regarded as unjust. Just
societies are therefore not necessarily “equal societies.”
There is no doubt that a society with a high general level of health, low rate of
violence and crime, mutual trust, and so on is a “good” society and also desirable. The
basic assumption of this book is that this can be achieved through eliminating differences
in status by reducing income differences—but it does not provide sufficient empirical
evidence for that assumption. On the basis of the data presented here, therefore, the
demand for income equality cannot be justified as a dictate of political acumen.
Finally, another fundamental question must be asked: is the social ideal that is the
implicit starting point and explicit goal of this book in fact desirable? This ideal becomes
clear when the authors write about the “better society.” Here they operate on the basis of
relatively clear-cut dichotomies. They contrast the negatively connoted “mass society”
with the positively connoted “community,” and discuss “social status” and “friendship”
as diametrically opposed principles of a social order: on the one hand mass society,
which is geared towards differences, power, privileged access to resources with no
consideration of others’ needs, hostility, exclusion, and social status, and on the other a
community based on friendship, balance, reciprocity, sharing, social responsibility, and
recognition of others’ needs. This is reminiscent of the culturally pessimistic models of
society discussed at the turn of the nineteenth to the twentieth century, which drew on the
duality of community and society in a similar way. At that time, too, communities were
considered to be the social forms that enabled people to coexist in solidarity on the basis
of a common background, long-term relationships, and full equality. Of course, at the
beginning of the twenty-first century we are somewhat the wiser from experience when it
comes to realizing this type of society oriented to the ideal of the community and the
associated consequences. Communities do not only provide warmth, identity, and
support; they are also exclusive—specifically of those who do not (or are not supposed
to) belong—and often ignore the need for individuality and the realization of fully
individual lifestyles. Therefore, it is perhaps precisely “society,” or in Wilkinson and
Pickett’s terminology “mass society”—which also allows impersonal relations for mutual
benefit in which competition not only burdens people but also spurs them on, and which
is based on the plurality of descent and diversity of lifestyles—that corresponds more
closely to the need for individuality and individual freedom. This is, of course, a question
to be answered normatively, not a matter to be decided by science.
