Inequalities in Health
THE SOCIAL CLASS DETERMINANTS OF INCOME
INEQUALITY AND SOCIAL COHESION
Carles Muntaner, John Lynch, and Gary L. Oates
The authors argue that Wilkinson’s model omits important variables (social
class) that make it vulnerable to biases due to model mis-specification.
Furthermore,the cultureof inequality
determinants of disease related to production (environmental and occupational
hazards) and the capacity of the relatively deprived for collective action. In
of violence has already been refuted. Shying away from social mechanisms
such as exploitation, workplace domination, or classist ideology might avoid
conflict but reduce the income inequality model to a set of useful, but simple
and wanting associations. Using a nonrecursive structural equation model that
tests for reciprocal effects, the authors show that working-class position is
negatively associated with social cohesion but positively associated with union
membership. Thus, current indicators of social cohesion use middle-class
standards for collective action that working-class communities are unlikely to
meet. An erroneous characterization of working-class communities as
noncohesive could be used to justify paternalistic or punitive social policies.
These criticisms should not detract from an acknowledgment of Wilkinson’s
investigations as a leading empirical contribution to reviving social
epidemiology at the end of the century.
I. FURTHER COMMENTS ON WILKINSON’S REPLY
What follows is several comments on Richard Wilkinson’s reply (1) to our
article (2) on his “income inequality and social cohesion” model (3). Wilkinson’s
reply spans a review of findings on income inequality and health, an expansion of
This work was supported by funds from NIH and CDC, project numbers Z01 MH02610-04,
Z01 MH02610-05, U48/CCU310821, and the Benedict Foundation (Dr. Muntaner).
This article is in two parts: part I by Drs. Muntaner and Lynch; part II by Drs. Muntaner, Oates,
International Journal of Health Services, Volume 29, Number 4, Pages 699–732, 1999
© 1999, Baywood Publishing Co., Inc.
his hypotheses on the relation between social cohesion and health, and a reply to
some of our criticisms. Because Wilkinson does not respond to a number of
issues we raised—for example, international dependency as determinant of
national income inequality; the role of political factors as determinants of income
inequality and population health; exploitation as explanation for income
inequalities; the need for formal definitions of social cohesion, functional forms
of its relation with health, and the exchange aspect of social cohesion (its
potential negative effects on health); the social psychology and the ethology of
social cohesion; the impact of income inequality among individuals in different
class locations; the policy implications of current research programs on social
cohesion; and the systemic approach to social inequality—we direct the reader to
the original article (2) and we concentrate here on Wilkinson’s reply.
Thus, we provide new arguments for incorporating social class in models of
social inequalities in health; we critically examine the support for the
psychological hypotheses that Wilkinson proposes as proximal determinants of
health; we challenge the notion that research on income inequalities per se
generates support for reducing social inequalities; and we provide a
philosophical framework that reveals some fundamental differences between the
“income inequality/social cohesion” model and other models of social
inequalities in health. Because Wilkinson criticizes us for not providing data1in
our analysis on the role of social class in his model (“a bald assertion that is
simply a matter of changing class relations” (1, p. 539)), we provide empirical
evidence in part II of this response. In part II we show that social cohesion is
shaped by class relations, and that Wilkinson’s measurement of social cohesion
excludes working-class forms of collective action.
Wilkinson’s reformulation of his income inequality, social cohesion, and
health model still does not explain the origins of income inequality (1). In
our previous critique (2) we proposed a series of economic and political
determinants of income inequality, and we addressed the public health
implications of placing this artificial boundary in the problem being analyzed.
Thus, omission of economic and political variables that might have an impact
on income inequality and population health may lead to biased estimates of
the aggregate relationship between income inequality and mortality rates. The
field of international studies, where social class and income inequality
700 / Muntaner, Lynch, and Oates
Epidemiologic pragmatism (e.g., as in “race” or “income” categories without hypotheses)
defends itself against realism (a synthesis of rationalism and empiricism) with accusations of “lack of
data.” However, more data on “race inequalities in health” do not provide any explanation and
precisely function as a deterrent of serious investigations on economic, political, and cultural racism
are considered predictors of democratization (rather than health), provides a
precedent for the current debate in public health. Following Bollen and
This omission [of direct measures of class structure] creates grave
problems in interpreting the coefficient estimate for income inequality.
The omitted variables are probably correlated with both income inequality
and political democracy, so the coefficient for income inequality is
biased. . . . Thus, although the coefficient estimate for income inequality
may indicate the operation of unspecified aspects of the class structure,
the exclusion of direct measures of the class structure means that we
cannot judge which particular classes play key roles in the process of
For example, even crude dichotomous class categories can account for 25 percent
of earnings inequalities as measured by the Theil Index (6).
A second methodological problem originates in Wilkinson’s interpretations of
these macro correlations (3). Ecological correlations of income inequality and
mortality or morbidity rates could emerge at the aggregate level even if income
inequality per se had no impact on individual mortality or morbidity risk (7).
Recently, though, with the incorporation of methods that take into account the
clustering of individuals into larger social units (8, 9), Sobaader (10) and
Kennedy and colleagues (11) have provided evidence that income inequality
indeed has an effect on individual health risk. However, in these studies, the
strength of the association between income inequality is lower than typically
reported by Wilkinson (3). This is in part due to the addition of aggregate (e.g.,
county poverty and absolute income) and individual level (e.g., income)
covariates (10) and to the fact that the strength of the relation between income
inequality and health depends on the level of aggregation (10). For example, the
effect of census tract income inequality on individual risk of anxiety disorders
can be weak because race and class segregation makes neighborhood incomes
homogeneous (12).2This is an instance where knowledge of political and
economic mechanisms, those that generate segregation above the neighborhood
level, is necessary to choose the appropriate level of aggregation to test for
income inequality effects.
Social Class and Income Inequality / 701
Multilevel analyses, though originally developed to estimate contextual effects, can be used to
emphasize the effects of individual behavior on health (e.g., 13). In fact, survey methods, because they
cannot measure the contextual nature of individual behavior itself, place the real limit on teasing apart
“individual” and “social context” contributions to individual health.
On “Misconceived Materialism,” “Outmoded Prejudices,”
and Other Red Scares
Wilkinson (1) states that our example on the recent increases in mortality in
the former Soviet Union (2) implies an endorsement of the centrally planned
economy that characterized that country before the “merchant capitalist” period
of the 1990s. We find this inference unjustified. Our point is that a sudden change
in the class structure (i.e., a political and economic change) had a major effect on
the health of its population (14). We did not deal with the potential causes of
declining population health since the mid-1960s, a phenomenon that we
acknowledged, however (2). Increased exploitation, which can exist in centrally
planned economies (1; see Economic Subsystem in 2, p. 73), could account for
this decline. The political oppression of labor organizations in the Soviet Union,
which Muntaner and Llorente (15) noted almost two decades ago, could also
account for that trend. In fact, because Wilkinson’s population health model calls
for reducing income inequality but leaves intact class relations, it gives implicit
support to any social system that reduces income inequalities regardless of
its class structure (e.g., the Soviet Union’s state socialism or the new example
(1, pp. 537–538)). On the other hand, the implications of a class exploitation
model (2) lead to the implicit rejection of both these social systems despite their
relative success in reducing income inequality. With regard to “misconceived
materialism” (1, p. 540), our example of a class system (2, Appendix) shows that
there is no manual/nonmanual labor distinction. Neither do we have any “out-
moded prejudice”3against psychosocial factors affecting health (16–18). The
issue is that psychosocial factors need to be integrated with social relations rather
than approached as isolated individual perceptions of relative inequality position
(17; see below).
Social Class: Relational and Stratified
Wilkinson states that class is mostly determined by income differences, mak-
ing class de facto an attitudinal response of individuals to their relative position
in the distribution of income. This framework does not specify where income
comes from, which is ultimately from production, as value is ultimately created
from labor, not the other way around (19). Similarly, the thought experiment pre-
sented by Wilkinson (1, p. 537) is unrealistic because he does not detail the rela-
tions of production in that society, only unequal distribution of incomes. As we
showed earlier (2, Appendix), a mere statement about lack of ownership (a legal
702 / Muntaner, Lynch, and Oates
Wilkinson's preoccupation with novelty often makes him overlook previous research. See the
section “Psychology in a Social Vacuum,” below.
criterion) does not describe the economic and political processes that take place
in the production of goods and services.
Contrary to Wilkinson’s interpretation, the presented class framework (2, p.
73) is not dichotomous (e.g., capitalist vs. worker) but integrates class relations
within a continuum. Muntaner and his collaborators have conducted several stud-
ies that show associations between relational/stratified class indicators and health
(e.g., wealthy capitalist, poor capitalist; large capitalist, small capitalist; worker,
manager; supervisor, worker) (4, 17, 20–22). Wilkinson is correct when he points
out that Boswell and Dixon’s rate of surplus value uses income (23). However,
theories of exploitation provide social mechanisms for the emergence of eco-
nomic inequality beyond income (e.g., wealth) and yield multiple measures that
are empirically distinct from income inequality (e.g., according to the role of
business services, depreciation, or benefits and pension plans in the calculation
of the rate of surplus value (24)). Therefore, the rate of surplus value is preferable
to catch-all income inequality indicators that do not specify social mechanisms
(24). Similarly, measures of “SES” (socioeconomic status) are strong predictors
of health outcomes but explain little (17), while its component indicators (occu-
pational stratification, income, educational credentials) are interrelated through
different social mechanisms (e.g., status attainment, inter- or intra-generational
mobility, age, labor markets, institutions) (eg., 25) and have unique effects on
PSYCHOLOGY IN A SOCIAL VACUUM:
WILKINSON’S CULTURE OF INEQUALITY
Wilkinson’s culture of inequality fulfills the urgent need for explaining the
micro effects of social inequalities. This is yet another instance where class anal-
ysis provides a more encompassing framework than the “income inequality and
social cohesion model.” The task of class analysis is precisely to understand not
only how macro structures (e.g., class relations at the national level) con-
strain micro processes (e.g., interpersonal behavior) but also how micro pro-
cesses (e.g., interpersonal behavior) can affect macro structures (e.g., via collec-
tive action) (27).
Wilkinson is certainly correct in addressing the lack of research on the psycho-
logical effects of inequality. While there is a substantial scholarship on the psy-
chology of racism and sexism, little research has been done on the effects of class
ideology (i.e., classism). This asymmetry could reflect that in most wealthy dem-
ocratic capitalist countries, income inequalities are perceived as legitimate while
gender and race inequalities are not (4). Most work on the psychology of inequal-
ity and classism has been qualitative (28–32). Most of these ethnographies, as
well as some recent empirical studies (33, 34), point to the relational aspects of
classism (e.g., the educated upper middle class holding views of inferiority about
the working class or its most deprived elements).
Social Class and Income Inequality / 703
But it is not only the psychological effect (e.g., humiliation) associated with
lower capacity to purchase goods and services that seems to matter. Thus,
although Wilkinson uses Sennett and Cobb’s classic The Hidden Injuries of Class
(29) for his argument about the psychology of inequality, he fails to mention
Sennett’s new volume (30), which stresses the erosion of control over the labor
process even among persons of relatively high income (e.g., the rise of nonstan-
dard work arrangements, lack of control due to mechanization). Attitudes about
the causes of social inequalities (i.e., classist ideology) are part of class relations.
Those with relatively high incomes (e.g., capitalists, managers, and profession-
als) hold attitudes that justify social inequalities, cast in terms of reductionist bio-
logical hypotheses (e.g., the inheritance of intelligence) or idealist lay psychol-
ogy (e.g., self, effort, morality, responsibility, will power) (32).
Wilkinson is correct in stressing the need to explain the social psychology of
health inequalities—that is, psychological social psychology (i.e., the study of
how society affects individual behavior and health)—as associations between
social indicators and mortality rates presuppose individual-level mechanisms
whereby social relations affect individual behavior and health. Nevertheless,
Wilkinson’s social psychology neglects precisely the impact of social (economic,
political, and cultural) relations on individual behavior. His approach is similar to
the U.S. interpersonal social psychology that was criticized by British psychol-
ogy more than 20 years ago (35, 36). That approach to social psychology was
abandoned because it focused on interpersonal behavior without analyzing the
social relations that determine it.
In addition to social psychology, Wilkinson uses the literature on animal and
human stress to summarize the evidence on the health effects of social inequali-
ties. While these fields have shown substantial progress in recent years (e.g.,
Sapolsky’s research on stress), Wilkinson’s presentation of them as “novelty”
does not acknowledge a long research tradition in the physiology of social hierar-
chies among primates (37) and in the study of human physiological psychology
(e.g., the Swedish school that linked responses of the sympathetic–adrenal
medullary system and the pituitary–adrenal cortical system to workplace stress
(38)). These omissions might contribute to overstating the degree of progress that
has been achieved with the “income inequality” approach and overlooking what
is known about the health effects of hierarchies within the labor process (38).
We next review several hypotheses advanced by Wilkinson that have already
been refuted decades ago in social psychology, or for which there is enough evi-
dence to cast doubt on their alleged generality.
The Relatively Deprived as “Men of Respect”:
A New Version of the Frustration/Aggression Hypothesis
Disrespect is now claimed to account for the association between inequality,
violence (i.e., lack of social cohesion), and health. But this sweeping generaliza-
704 / Muntaner, Lynch, and Oates
NOTOPIN “Parties are only interested in people’s votes, but not their
Accompanying response categories are 1 = strongly agree; 2 = agree; 3 = dis-
agree; 4 = strongly disagree.
“other political organizations”
“church or religious organizations”
“racial or ethnic organizations”
“special interest groups or hobbies”
For each item, membership is coded 1 and non-membership, 0.
Membership is coded 1; non-membership, 0.
1 = Working class—employees who fall within none of the four middle- and up-
per-class occupation categories (see below)
0 = Middle and upper class—a combination of the following occupational cate-
capitalists: self-employed and supervising at least one person
self-employed: self-employed and supervising no one
supervisors: employed individuals who supervise others
professional/managerial: non-supervisory employed individuals who hold/
held jobs classified as such by the International Labor Organization.
Note: Classifications are based on respondents current or last job. Respondents
who had never held a paid job were excluded from the analysis.
Z of family income
Standardized (gross yearly) family income in dollars
Respondent’s education in years
728 / Muntaner, Lynch, and Oates
Construction of the Latent Variables
Latent variables were computed by summing the product of each item by its
weight. Weights were obtained by dividing the factor score regression coefficient
associated with each item by the sum of all factor scores (for items in each scale)
coefficients. Each weight represents the indicator’s proportional contribution to
the overall scale. The indicators of conventional participation do not all have
identical metrics, and so were standardized prior to being combined.
Social Class and Income Inequality / 729
Factor loadingsa,bfor each indicator in the four latent variables of the model:
conventional participation, legal protest, political efficacy, and organization membership
Reciprocal effects model
Time 1Time 2
aP < .01 or < .05 for each loading; — signals the dependent variables in the non-recursive models.
bLoadings suggest that latent variables are successfully measured with loadings exceeding .35 in
the overwhelming majority of cases.
Loadings for indicators of the latent variables included in the preliminary and
reciprocal effects models are listed in Table 4.
REFERENCES (PART II)
1. Wilkinson, R. G. Unhealthy Societies: The Afflictions of Inequality. Routledge,
2. Muntaner, C., and Lynch, J. Income inequality, social cohesion, and class relations:
A critique of Wilkinson’s neo-Durkheimian research program. Int. J. Health Serv.
29: 59–82, 1999.
3. Kawachi, I., et al. Social capital, income inequality, and mortality. Am. J. Public
Health 87: 1491–1499, 1997.
4. Anastasi, A. Psychological Testing. Macmillan, New York, 1982.
5. Fantasia, R. Cultures of Solidarity: Consciousness, Action and Contemporary Ameri-
can Workers. University of California press, Berkeley, 1988.
6. McNall, S. G., Levine, R., and Fantasia, R. Bringing Class Back In: Contemporary
and Historical Perspectives. Westview Press, Boulder, Colo., 1991.
7. Johnson, J. V., and Johansson, G. The Psychosocial Work Environment: Work
Organization, Democratization and Health. Baywood, Amityville, N.Y., 1991.
8. Verba, S. K., Lehman Schlozman, K., and Brady, H. E. Voice and Equality: Civic Vol-
untarism in American Politics. Harvard University Press, Cambridge, Mass., 1995.
9. Wallace, M., and Jenkins, J. C. The new class, postindustrialism, and neocorporatism:
Three images of social protest in western democracies. In The Politics of Social Pro-
test: Comparative Perspectives on States and Social Movements, edited by J. C.
Jenkins and B. Klandermans, pp. 96–137. University of Minnesota Press, Minneapo-
10. Morris, A. D., and Mueller, C. M. Frontiers in Social Movement Theory. Yale Univer-
sity Press, New Haven, 1992.
11. Przeworski, A. Capitalism and Social Democracy. Cambridge University Press,
New York, 1985.
12. Przeworski, A. Class, production and politics: A reply to Burawoy. Socialist Rev.
2: 87–111, 1989.
13. Roemer, J. A General Theory of Exploitation and Class. Harvard University Press,
Cambridge, Mass., 1982.
14. Wright, E. O. Classes. Verso, London, 1985.
15. Coleman, J. Foundations of Social Theory. Harvard University Press, Cambridge,
16. Burt, R. S. Toward a Structural Theory of Action. Academic Press, New York, 1982.
17. Becker, G. S. The Economic Approach to Human Behavior. University of Chicago
Press, Chicago, 1976.
18. Elster, J. Solomonic Judgments: Studies in the Limitations of Rationality. Cambridge
University Press, New York, 1989.
19. Cook, K. S., and Levi, M. The Limits of Rationality. University of Chicago Press, Chi-
20. Macy, M. W. Learning theory and the logic of critical mass. Am. Sociol. Rev. 55:
730 / Muntaner, Lynch, and Oates
21. Rosenberg, A. Economics—Mathematical Politics or Science of Diminishing Re-
turns? University of Chicago Press, Chicago, 1992.
22. Zey, M. Decision Making: Alternatives to Rational Choice Models. Sage, Newbury
Park, Calif., 1992.
23. Esser, H. How “rational” is the choice of “rational choice”? Rationality and Society
5: 408–414, 1993.
24. Petersen, T. On the promise of game theory in sociology. Contemp. Sociol. 23:
25. Herrnstein, R. J., and Prelec, D. Melioration. In Choice Over Time, edited by
G. Lowenstein and J. Elster. Rusell Sage Foundation, New York, 1992.
26. Goldstone, J. A., and Opps, K. D. Rationality, revolution, and 1989 in Eastern Europe.
Rationality and Society 6(1): 5–7, 1994.
27. Bobo, L., and Gilliam, F. Race, sociopolitical participation, and black empowerment.
Am. Polit. Sci. Rev. 84: 377–393, 1990.
28. Guterbock, T. M., and London, B. Race, political orientation, and participation: An
empirical analysis of four competing theories. Am. Sociol. Rev. 48: 439–453, 1983.
29. Paulsen, R. D. Education, social class and participation in collective action. Sociol.
Educ. 64: 96–110, 1991.
30. Sherkat, D. E., and Blocker, T. J. The political development of the sixties’ activists:
Identifying the influence of class, gender, and socialization on protest participation.
Social Forces 72: 821–842, 1994.
31. Finkel, S. Reciprocal effects of participation and political efficacy: A panel analysis.
Am. J. Polit. Sci. 29: 891–913, 1985.
32. Herring, C. Splitting the Middle: Political Alienation, Acquiescence, and Activism
Among America’s Middle Layers. Praeger, New York, 1989.
33. Neal, A. G., and Seeman, M. Organizations and powerlessness: Test of the mediation
hypothesis. Am. Sociol. Rev. 29: 16–26, 1964.
34. Almond, G., and Verba, S. The Civic Culture. Little Brown, Boston, 1963.
35. Gamson, W. A. Power and Discontent. Dorsey Press, Homewood, Ill., 1968.
36. Gould, R. Collective action and network structure. Am. Sociol. Rev. 58(2): 182–196,
37. Campbell, A., Gurin, G. and Miller, E. The Voter Decides. Row Peterson, Evanston,
38. Weakliem, D. L., and Heath A. F. Rational choice and class voting. Rationality and
Society 6(2): 243–270, 1994.
39. Zentral Archive Fuer Empirische Socialforschung. Political Action: An Eight Nation
Study—1974. Political Action, Cologne, Germany, 1974.
40. Zentral Archive Fuer Empirische Socialforschung. Political Action: An Eight Nation
Study—1981. Political Action II, Cologne, Germany, 1981.
41. Joreskog, K. G., and Sorbom, D. Lisrel 8 User’s Reference Guide. Scientific Software
International, Chicago, 1993.
42. Oates, G. L. Self-esteem enhancement through fertility? Socioeconomic prospects,
gender and mutual influence. Am. Sociol. Rev. 62: 965–973, 1997.
43. Oliver, P. E., and Marwell, G. Mobilizing technologies for collective action. In
Frontiers in Social Movement Theory, edited by A. D. Morris and C. McClurg
Mueller, pp. 251–272. Yale University Press, New Haven, 1992.
Social Class and Income Inequality / 731
44. Navarro, V. The Politics of Health Policy. Baywood, Amityville, N.Y., 1994.
45. Wright, E. O. Class Counts: Comparative Studies in Class Analysis. Cambridge Uni-
versity Press, New York, 1997.
46. Marshall, G., et al. Classes in Modern Britain. Hutchinson, London, 1988.
47. Browne, M. W., and Cudeck, R. Alternative ways of assessing mode fit. In Testing
Structural Equations Models, edited by K. A. Bollen and J. S. Long. Sage, Newbury
Park, Calif. 1993.
48. Fox Piven, F., and Cloward, R. A. The Breaking of the American Social Compact. Vin-
tage Books, New York, 1999.
49. Crossing the Rubicon on welfare. Economist, March 6, 1999, pp. 53–54.
50. Cress, D. M., and Snow, D. A. Resources, benefactors, and the viability of homeless
SMOs. Am. Sociol. Rev. 61: 1089–1109, 1996.
51. McCarthy, J. D., and Wolfson, M. Resource mobilization by local social movement
organizations. Am. Sociol. Rev. 61: 1070–1088, 1996.
52. Opp, K.-D. and Gern, C. Dissident groups, personal networks, and spontaneous coop-
eration: The East German revolution of 1989. Am. Sociol. Rev. 58: 659–680, 1993.
53. Schwartz, J. Functional explanation of metaphysical individualism. Sci. Society 17:
54. Muntaner, C., and O’Campo, P. A critical appraisal of the demand control model of
the psychosocial work environment: Epistemological, social, behavioral and class
considerations. Soc. Sci. Med. 36: 1509–1517, 1993.
55. Diez-Roux, A. V. Bringing context back into epidemiology: Variables and fallacies
in multilevel analysis. Am. J. Public Health 88(2): 216–222, 1998.
56. Burawoy, M. The Politics of Production. Verso, London, 1985.
57. Wright, E. O. Rethinking, once again, the concept of class structure. In The Debate
on Classes, edited by E. O. Wright, et al. pp. 269–348. Verso, London, 1989.
58. Wright, E. O. Typologies, scales and class analysis: A comment on Hallaby and
Weakliem. Am. Sociol. Rev. 58: 35–36, 1993.
59. Muntaner, C., and Stormes, J. Social class and behavior. Psychol. Rep. 79: 379–382,
Direct reprint requests to:
Dr. Carles Muntaner
Institute of Occupational and Environmental Health
West Virginia University School of Medicine
3801 Health Sciences South
P.O. Box 9190
Morgantown, WV 26506-9190
732 / Muntaner, Lynch, and Oates