Ethical principles dictate that all people should be valued equally and have equal opportunities to be healthy. Health is essential for well-being and economic self-sufficiency; thus, resources needed for health should be distributed equitably, which many have interpreted to mean according to need. It is difficult, however, to define and measure need (3,9,10).

Human rights principles can provide assistance. According to well-established human rights norms and agreements signed by almost all nations, the right to health is the right to attain the highest level of health possible (11-13). This concept has been criticized for being difficult to operationalize for measurement purposes (14). However, the health of the most privileged social groups (eg, the dominant racial/ethnic group or the wealthiest group) indicates what should be possible for everyone (3,15). Human rights norms, principles, and agreements oblige governments to make good faith efforts toward progressively removing obstacles to all people's realizing their full health potential, particularly for those social groups that historically have experienced more obstacles.

Differences in a range of health indicators according to diverse measures of social class have been observed for centuries (16) and across virtually all societies in which they have been studied (17). Health has been reported by race in the United States for more than a century (18,19), although the issue of racial/ethnic disparities in health became prominent only during the past 2 decades (20,21). Many health researchers routinely report results by racial/ethnic group but do not examine health differences by markers of social class, such as income or education. Inadequate socioeconomic information often has reinforced a widespread tendency to make unfounded assumptions about the nature of racial or ethnic disparities in health, reifying genetic or "cultural" explanations and deflecting attention from potentially remediable social factors (22).

A disparities lens can focus on neglected factors in health research, increasing the likelihood of sound conclusions, not only about social disparities in health but also about a range of research questions, including ones assumed to be purely biomedical. Failure to adequately consider social factors can result in erroneous conclusions from research findings, even on many questions that are not in themselves focused on social issues but for which social factors may play a role as confounders, mediators, or effect-modifiers (23).

Social disadvantage can be based on material conditions, determined by access to resources and services that affect health such as adequate nutrition, sanitation, housing, and medical care. It also can be of a psychosocial nature, based on human relationships and their psychological effects. For example, unfair treatment based on one's race or ethnic group can cause psychological distress. In addition, one's awareness of being in a group that has historically suffered discrimination could act as a chronic stressor, even in the absence of overt incidents of unfair treatment. These dimensions often coexist and interact. Material disadvantage (eg, resulting from inadequate income or wealth) can affect obesity by influencing the ability to purchase nutritious food or to live in a neighborhood with safe, pleasant places to exercise and markets that sell affordable fresh produce. Material hardship also could increase obesity risk insofar as it is a source of chronic stress; stress could limit people's ability to change weight-related behaviors even when informed and motivated (24-26). Low educational attainment could increase the risk of obesity by limiting economic opportunities or one's ability to understand and act on health information.

Racial or ethnic group is closely associated with social advantage and disadvantage and with health disparities. Although each broad racial or ethnic group is heterogeneous, overall, blacks, Hispanics, and American Indians have the lowest and Asian Americans the highest incomes and educational levels; whites have intermediate levels (27-29). Racial or ethnic differences thus often reflect socioeconomic differences, which can affect health through material pathways (23).

Experiences of racism could include not only overt incidents of intentional discrimination but also experiences in which unintended harm results because of deeply rooted structural arrangements, such as those perpetuating racial residential segregation. Racial segregation systematically deprives blacks and Hispanics of opportunities to live in health-promoting neighborhoods, in part by constraining their economic opportunities (5,30). These experiences could deleteriously affect health outcomes both through material pathways and through psychosocial pathways involving stress and physiologic responses to stress related to awareness of unfair treatment or stigmatization (as a member of a socially excluded group). Recent advances in understanding the neurophysiology of stress and its effects on chronic disease have greatly increased our ability to understand how both material and psychosocial disadvantage can harm health (31).

Any condition associated with stigma or lower social acceptance — such as obesity — could lead to social disadvantage and accompanying adverse health effects that are not intrinsic to that condition. Adverse health effects could occur through material or psychosocial pathways. Examples include physical or mental disability, HIV infection, or other highly stigmatized diseases. Similarly, nonheterosexual orientation can result in discrimination or social exclusion, putting one's health at risk in multiple ways. These experiences of discrimination are rarely measured in health studies.

Time is another dimension of social advantage and disadvantage that can be crucial to understanding health disparities. It seems likely that not only the depth but also the duration of exposure to disadvantage could matter greatly for health. Exposures that are potentially obesigenic, such as a high-calorie diet, a crime-infested neighborhood without safe places to exercise or play, or a resource-strapped school that offers children few opportunities for supervised exercise, will likely have a larger effect given a longer duration. Yet even when these factors are measured at all, time is rarely considered. Current or last year's income may be measured but generally not whether a person was poor as a child. This oversight tends to underestimate racial/ethnic disparities in social advantage. Our research with population-based data on postpartum women in California confirms that at each level of current income or education, non-Hispanic black and Hispanic women are more likely than their non-Hispanic white counterparts to have grown up in households of lower socioeconomic status (as reflected by their parents' educational attainment) (23).

A disparities perspective leads us to ask about not only the antecedents of disparities in obesity but also the differential consequences of obesity for people in different social groups. Finn Diderichsen of the Karolinska Institute in Stockholm has developed a schematic diagram highlighting the dynamic nature of how health disparities are produced and reproduced over time. Figure 2, adapted from Diderichsen, depicts how social position or stratification (the extent to which different groups are sorted into hierarchies of wealth, influence, and opportunities) leads to different health-promoting or health-damaging exposures for different social groups. Differences in social position influence not only whether a person is exposed to a given health risk but also differential vulnerability to disease incidence and severity and subsequent social consequences of illness. For example, a highly educated person who, because of obesity, develops heart disease with activity limitations may be less likely than a manual worker with little schooling to become unemployed. The highly educated person is more likely than the manual worker to have work that is knowledge-based, less affected by physical capacity, and more easily performed at home or on a more flexible schedule.

Diderichsen's diagram calls attention to multiple levels at which pathways toward health disparity can be interrupted by policies, from the most proximal level (proximal or downstream in relation to the outcome; eg, medical treatment ameliorating the health damage done by harmful exposures without addressing the exposures themselves) to the most distal level (policies in the social context that may blunt the degree of social stratification, such as policies supporting universal high-quality education beginning in early childhood, and poverty reduction).