Estimating the stability of census-based racial/ethnic classifications: The case of Brazil

Abstract

This study presents a method of estimating the degree to which people change their racial/ethnic identity from one census enumeration to another. The technique is applied to the classification of skin colour in Brazil (white, black, brown, yellow). For the period 1950-80, the findings show a deficit of 38 per cent in the black category and a gain of 34 per cent in the brown category, suggesting that a large proportion of individuals who declared themselves black in 1950 reclassified themselves as brown in 1980. Estimates for 1980-90, adjusted for the effects of international migration, reveal a similar pattern, although the magnitude of colour reclassification may have declined somewhat during the 1980s. Procedures to determine the stability of racial/ethnic identity produce data useful to recent policy initiatives that rely on demographic censuses to measure changes in the status of minority groups in less developed countries.

A resurgence of black identity in Brazil? Evidence from an analysis of recent censuses

“…This surplus implies that at least 2.2 million people in 2000 were newly reclassified blacks. In relative terms, this represents a net increase of 33% due to racial reclassification alone, which contrasts with the net decrease of 4% experienced by the black category in the 1980s, as documented by Carvalho et al (2004). The estimates for the period between 2000 and 2010 show that the number of newly reclassified blacks continued to grow.…”

“…11 Since the race question was not asked in the 1970 census, a mean decadal reclassification rate for the 1960s and 1970s was obtained by dividing the estimated rate for the 1960-80 period by two. This is a simplification of the more complex procedure proposed by Carvalho et al (2004) to convert a rate of reclassification to a decadal basis. Nonetheless, separate calculations showed that their more elaborate technique 11 The child survival method and the model life tables presented by Frias and Rodrigues (1980) were used to estimate intercensal survival rates, following the same method described in section 4 for the 1990s and 2000s.…”

“…Using data from the 1950, 1980 censuses, these studies found a clear pattern of reclassification: while the number of people selfidentifying as black or white declined, the number of people in the brown category increased (Carvalho et al 2004;Wood 1991;Wood and Carvalho 1994).…”

“…For instance, based on the mortality and migration rates prevailing in the period covered by their study, Carvalho et al (2004) estimated that 10.2 million people who self-identified as brown in the 1950 census were expected to survive to the 1980 census, when this population reached ages 30 and older. However, 13.6 million people ages 30 and older actually identified as brown in the 1980 count.…”

“…While the proportion of the population self-identifying as "brown" increased from 26% to 43% between 1950 and 1991, the proportion self-identifying as "black" or "white" declined from 11% to 5% and from 62% to 52%, respectively. Ruling out differentials in fertility, mortality, or migration rates as potential explanations, previous studies have established racial reclassification as a driving force behind this trend: over their life course a significant number of people changed their response in the censuses from black or white to brown (Carvalho, Wood, and Andrade 2004;Wood and Carvalho 1994). This pattern is consistent with the 'racial democracy' narrative.…”