April 2006
Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29

Authors' Committee


Matt Blackwell (Gov)


Martin Andersen (HealthPol)
Kevin Bartz (Stats)
Deirdre Bloome (Social Policy)
John Graves (HealthPol)
Rich Nielsen (Gov)
Maya Sen (Gov)
Gary King (Gov)

Weekly Research Workshop Sponsors

Alberto Abadie, Lee Fleming, Adam Glynn, Guido Imbens, Gary King, Arthur Spirling, Jamie Robins, Don Rubin, Chris Winship

Weekly Workshop Schedule

Recent Comments

Recent Entries



SMR Blog
Brad DeLong
Cognitive Daily
Complexity & Social Networks
Developing Intelligence
The Education Wonks
Empirical Legal Studies
Free Exchange
Health Care Economist
Junk Charts
Language Log
Law & Econ Prof Blog
Machine Learning (Theory)
Marginal Revolution
Mixing Memory
Mystery Pollster
New Economist
Political Arithmetik
Political Science Methods
Pure Pedantry
Science & Law Blog
Simon Jackman
Social Science++
Statistical modeling, causal inference, and social science



Powered by
Movable Type 4.24-en

« April 10, 2006 | Main | April 12, 2006 »

11 April 2006

Unstable Racial Identities

Felix Elwert

Race is a surprisingly malleable construct, though it’s usually taken as fixed in statistical models. In a recent paper with Nicholas Christakis (Widowhood and Race, American Sociological Review Vol 71(1), 2006) I had to engage changing racial responses head on.

Assorted previous research has shown that people may change their racial self-description over time because they are multiracial, when they marry somebody of a different racial group, or – not to be neglected – because the answer choices in surveys may change over time.

Most people think that unstable or changing racial self-identification is an issue largely confined to a small group of multiracial individuals. This is a country, after all, of the one-drop rule. But research, including our own, shows that that isn’t so.

In a supplementary analysis of the 2001 Census Quality Survey (CQI), we showed that the racial self-identification of “whites" is also surprisingly unstable. The CQI asks more than 50,000 respondents twice within the span of just a few months to identify their own race. Once they were allowed to select only one race, and the other time they were given the option of selecting multiple races (this gets at the difference between the old and the new Census race questions). The answers were then matched to individual responses from the official 2000 Census.

Depending on whether we compared between consecutive responses to the same race question on the Census and the CQS, or between the different questions asked in the two waves of the CQS, and whether we treat “Hispanic" as a category distinct from black and white, the agreement between answers for whites ranged from 95.6 to 97.5. We obtained really similar answers for blacks.

Meaning, between 2 and 5 percent of people who used to identify as white, would call themselves either something else or a mixture of races when given the chance. And the percentage of “whites" who will change their racial self-description as a function of question wording is about the same as the percentage of “blacks" who will do likewise.

Posted by Felix Elwert at 6:00 AM