Abstract:
|
While studies that use race as a variable tend to treat it as constant over time, much recent work has suggested that individuals can change their racial identification. Leibler et al examine matched 2000 and 2010 Census data and describe the frequency of changes in race reporting by the characteristics of the individuals involved. This paper builds on that work, using the same data to fit logistic regression models to estimate the probability that an individual would change racial identification between the 2000 and 2010 Censuses based on that individual’s characteristics. I fit both an overall model and separate models for each race group, and analyze the differences between the race groups in the contributions individual characteristics make to the probability of change. I also investigate the degree to which the census-to-census changes may understate the true probability of race change by using the census-to-census change probabilities to estimate lower bounds for the probabilities that individuals may change racial identification more than once in the census-to-census period.
|