Race/ethnic status in the 2004 SCF

Sherman D. Hanna, October 13, 2006

Note that the new Hispanic variable in the 2004 SCF allows for different approaches to defining racial-ethnic self-identification by the respondent in each household.

See Endnote 1 in a forthcoming article in the Journal of Consumer Affairs (2007): http://ehe.osu.edu/cs/scf/scf-critique-lindamood-hanna-bi.pdf

Note also that as Jie Qin of the University of Missouri-Columbia pointed out in 1997 (/imp.htm)

a particular household might be listed as different racial-ethnic groups in different implicates.

In our analysis of the 2004 SCF:

With the old one question variable, the number of white respondents ranged from 3517 to 3521 in different implicates, and the number of Black respondents ranged from 483 to 485.

With the new combined racial/ethnic variable, the number of non-Hispanic whites ranged from 3439 to 3443.

(As noted in endnote 1 of Lindamood, Hanna and Bi, researchers using the public datasets before 2004 should not refer to non-Hispanic whites, because the choice of "white" in the old one question variable does not mean that the respondent did not also consider "Hispanic" as an additional identification.  The Federal Reserve Board staff with access to the internal datasets can properly refer to non-Hispanic whites, because they can identify which respondents chose Hispanic as a second choice.)

See SAS code below.

*NOTE: prior to 1998, the SCF only asked for one response.

In 1998, respondents were allowed to give multiple responses,

but they were asked to give first the category they identified

with most strongly. Few people gave more than one response.

For purposes of continuity with prior data, define the 1998+

variable in terms of the strongest identification;

*beginning in 2004, respondents were also asked a question to

determine whether they were of Hispanic/Latino culture or origin;

* 1=white non-Hispanic, 2=black/African-American, 3=Hispanic,4=Asian (only available in internal data set, see codebook),5=other;

IF X6809=1 THEN RACE=1;
ELSE IF X6809=2 THEN RACE=2;
ELSE IF X6809=3 THEN RACE=3;
ELSE RACE=4;

IF X6809=1 & X7004^=1 THEN H_RACE=1;
ELSE IF X6809=2 & X7004^=1 THEN H_RACE=2;
ELSE IF X6809=3 | X7004=1 THEN H_RACE=3;
ELSE H_RACE=4;