Census Researchers Pretty Much Conclude NYTimes Messed Up

Jun 12, 2014
6:04 PM

So about a month since Nate Cohn of The New York Times boldly proclaimed that more Latinos were becoming “white” after basing his “analysis” on a partial report chronicled by a Pew senior research reporter, the actual authors of the study Cohn had referenced finally offered a more specific clarification about their initial study. Here is an excerpt of what these co-authors published today on the US Census’ Research Matters blog (phrases in bold italics our emphasis):

Via New York Times

Via New York Times

Our preliminary findings have generated a lot of interest, and we are taking this opportunity to clarify what we have found so far.

Previous research (see references in abstract) suggests that responses to questions of race and Hispanic origin can change over time or across contexts for a variety of reasons, including life experiences, changing social forces, a change in who is reporting the race/Hispanic origin, or questionnaire design.

The race and Hispanic origin response changes that we observe from 2000 to 2010 may be influenced by many of these factors, and some groups may be differentially impacted by questionnaire design changes from Census 2000 to the 2010 Census.

Our research paper documents the extent of the response changes and gives information about their patterns. After these patterns are well-documented, further research can be done about why responses are changing.

Our research used census responses that were linked to the same anonymized individual in both 2000 and 2010.  While this yields a rich and unique dataset, it is not representative of the total United States population.

Like many other demographers, sociologists, and census researchers, we found that some race and Hispanic origin responses can and do change between censuses.  In preliminary results, we found that race and Hispanic origin responses changed for about 10 million people (or 6 percent) out of the 168 million individuals in our linked dataset from Census 2000 to the 2010 Census.

Race response change was concentrated in some groups, specifically American Indians and Alaska Natives, Native Hawaiians and Other Pacific Islanders, people who reported multiple races, and Hispanics who reported a race. Responses were generally stable among single race non-Hispanic whites, blacks, and Asians in our data

The most common response change was from Some Other Race (SOR) to single race white among those who identified (or were identified by someone in the household) as Hispanic in both Census 2000 and the 2010 Census. The second most common response change was the inverse — from single race white to SOR for those who reported (or were reported as) Hispanic in 2000 and 2010.

Across most types of response change, we note that “inflows” to each race/Hispanic group are similar in size to the “outflows” from the same group, indicating that cross-sectional data might show a small net change. We also note that race and/or Hispanic origin response change occurs in a wide variety of ways. These included changes from multiple-race to single-race, from single-race to multiple-race, from one single race to another, or adding (or dropping) a Hispanic response.

Again, these findings are preliminary and cannot be considered final. We are still analyzing the data and plan to release more detailed results in a working paper in the coming months. We caution readers not to reach conclusions based solely on the preliminary research.  We look forward to talking about the final results when the process is complete.