There has been some controversy over whether teachers should be required to record the "race" or "population group" of pupils, as a means of measuring and monitoring transformation in schools.

Debate over the recording of "race' or "ethnic identity" in surveys and censuses is not peculiar to South Africa. A search of the internet with the parameters "census+race" found nearly 8.3 million entries.

Literature on race in American censuses is particularly extensive. A description of a recent book by Peter Skerry - Counting on the Census? Race, Group Identity and the Evasion of Politics - notes that "Since the US constitution first instructed that a slave be counted as only three-fifths of a person, the census has been caught up in America's racial dilemmas.

Today it is torn by controversies over affirmative action, evolving racial identities, and minority undercounts."

In the US, the 2000 population census questionnaire provided more options under ethnicity than in previous years, and for the first time allowed people to identify themselves with more than one race. The census questions relied on self-identification.

For example, some "Latinos" consider themselves white, or Caucasian. Others say they are mestizo, or mixed race (of Spanish and Indian descent). Still others identify themselves as black or Asian.

Members of the Mexica movement in Los Angeles and others with strong indigenous roots prefer to use Indian or Native American as their race. As in South Africa, there are those who hate all of the racial or ethnic modifiers and want to be known solely as citizens of their country.

In South Africa, intense discussion preceded the decision to include "race" or "population group" as a variable in the collection of official statistics. Some argued that the non-racial aims of the majority of formations in the anti-apartheid struggle rendered categorisation by race offensive and unnecessary.

Tony Holiday, a philosophy lecturer at the University of Western Cape and an underground activist for the SA Communist Party jailed in the 1970s, recalls a secret briefing with Joe Slovo, who felt that it might be best to make it illegal to refer to a person's race for at least a period after apartheid.

Others pointed out that the repeal of the Population Registration Act in 1991 removed any legal basis for specifying "race".

The Identification Act of 1997 makes no mention of race. On the other hand, the Employment Equity Act speaks of "designated groups" being "black people, women and people with disabilities". The Act defines "black" as referring to "Africans, coloureds and Indians".

Apartheid and the racial identification which underpinned it explicitly linked race with differential access to resources and power. If the post-apartheid order was committed to remedying this, race would have to be included in surveys and censuses, so that progress in eradicating the consequences of apartheid could be measured and monitored.

This was the reasoning that led to a "self-identifying" question about "race" or "population group" in both the 1996 and 2001 population censuses, and in Statistics SA's household survey programme.

There were, of course, those who chose not to specify a racial identity: provision was made for this in the category "unspecified/other". However, the vast majority of South Africans enumerated in, for example, Census 2001, completed the question on race or population group.

Analysis of data from Census 2001 shows that association with factors of underdevelopment is still overwhelmingly attached to people who classify themselves as African.

These factors include level of labour skill, income, access to telephones, electricity, sewerage and water, and standard and density of accommodation. Factors associated with development increase as they are linked to people who self-classify as coloured, Indian and finally white.

The association with groups of factors used to create social, economic and development indicators is distributed predominantly - although not exclusively - by race. For the vast majority of people in South Africa, race remains an important indicator of access to resources.

Without some reference to a historic notion of race, it is not possible to measure and monitor progress in eradicating the consequences of apartheid. It is, however, open to debate whether race as a variable in statistical collection and analysis is adequate on its own.

It is for this reason that Stats SA incorporates other factors in the derivation of social, economic and development indicators. In his way, we are attempting to develop, to use Tony Holiday's words again, "different, more fine-grained ways of measuring disadvantage".

n Pali Lehohla is South Africa's statistician-general and head of Stats SA. For more information on Stats SA, visit www.statssa.gov.za, or contact user services on (012) 310-8600. Tony Holiday's comments first appeared in the Cape Times, August 11 (Death of race classification overdue)