Self-classification occurs anytime an individual reports an identity, characteristic, or membership within a social group. Though the use of self-classification as a discrete form of measurement coincides with the twentieth-century popularization of social surveys in Britain and the United States (Gordon 1973), the origins of classification have roots in antiquity. Efforts to systematically distinguish plants from animals began with Aristotle, and following the influential work of the Swedish botanist Carolus Linnaeus (1707—1778), classification became the scientific standard for expressing organizational hierarchies in the natural world (Frangsmyr, Lindroth, Eriksson, and Broberg 1983). Taxonomic classification groups individual units by the number of properties they share, with broad, loosely related parent categories subdivided into more detailed, closely related subcategories. Societies are also organized hierarchically, and to the extent that population groups differ on outcomes of interest to social scientists, classification can be viewed as a prerequisite of all comparative research and studies of inequality in particular.

In the natural sciences units are classified by the investigator, and taxonomies are assumed to represent objective groupings based on universal laws. Social taxonomies are embedded within social relations, however, and cannot be viewed as objective or universal (Durkheim [1912] 1965). Subjects often classify themselves in social research, and many surveys, polls, and population censuses are composed entirely of self-reported data.

In many ways self-classification has revolutionized social research, particularly in areas of attitude and opinion polling. Self-reports of personal well-being, self-esteem, and presidential approval are common examples of types of subjective measures widely used in cross-sectional comparisons and trend analyses. Even objective information such as height and weight can be self-classified without incurring the costs of direct measurement.

Because self-reports are inherently subjective, however, they sometimes provide inaccurate or unreliable measures of objective characteristics. People may lack relevant knowledge about their personal histories; even seemingly straightforward measures, such as the place and timing of birth, are unknown in certain instances. In other cases, individuals may misrepresent themselves, particularly on measures that are perceived to be intrusive or potentially stigmatizing, as with measures of income, substance abuse, and even height or weight (Strauss 1999). Although advances in questionnaire design and interview techniques have improved the quality of self-reported data, there are limits to what these advances may achieve. For example, many individuals are unwilling to self-classify as perpetrators of discrimination on survey questionnaires, which has spurred the development of "deception designs," such as housing and job search audits, that can be used to circumvent respondents’ reluctance to admit socially undesirable behavior (Yinger 1986).

Discrepancies between self-classification and direct measures are also a risk when the constructs underlying those measures reflect a combination of objective and subjective realities. Many axes of inequality fit this criterion. Though concepts such as race, class, and gender can be reduced to ostensibly objective terms (ancestry, income, and sex, respectively), these simplifications ignore the social embeddedness of identity, which is shaped by the dynamic interplay between social actors, institutional forces, and historical contingencies. Systems of racial hierarchy evolved quite differently in North and South America (Graham 1990), for example, so it is plausible for similar-looking individuals from each region to self-classify in different ways. Because identities reflect the social construction of race in each region, however, it is inappropriate to view one identity as more accurate than the other, much less to justify the use of "objective" criteria, such as ancestral descent, which may have little bearing on the social consequences of race in these regions.

Next post:

Previous post: