The American People: Census 2000 (review)

The poverty rate for the elderly has decreased from 37 percent in 1959 to only 7 percent in 1999(62). In 2000, only 56 percent of all U.S. children were living with both biological parents (213). Most single-parent families with children are still female-headed: only 2.1 percent were headed by men in 2000 (182). With 53 percent, Miami has the highest share of foreign-born residents among metropolitan areas compared to a mere 3 percent in Pittsburgh and St. Louis (278). Less than 3 percent of American population describes themselves with more than one race (335).

We learn these and many other interesting facts from The American People: Census 2000, edited by Reynolds Farley and John Haaga. It is a compilation of articles on economic inequality, trends in marriage and family, gender disparities, historical cohorts, immigration, and racial composition in the United States. The authors range from the former head of the U.S. Census Bureau, Kenneth Prewitt, to well-known life-course specialists including Mary Hughes and Angela O'Rand.

The authors use simple-to-read statistical analysis (cross-tabulations and percentages) making it an edition that should and could be read both by sociologists but also by a much wider audience. The textbook style and boxes with examples and definitions make it an enjoyable read. All chapters end with a discussion of causes of changes. Thus, the book would be a good one to use in any introduction to population studies course.

In addition to addressing the changes in U.S. population, the book focuses on the changes in census data collection. Specifically, it emphasizes the importance of the census long form and the new American Community Survey, which is replacing the traditional census long form. It is one of the last books written from decennial census; the next ones, in comparison, will employ the annual ACS. Already now many authors have used additional, more frequently collected data sources than the old census long form (Current Population Surveys, General Social Survey, Vital Statistics Reports, Surveys of Income and Program Participation). This shows that continuous data collection is definitely welcome and the decennial census alone as a source is not always sufficient.

The book highlights the potential difficulties in using the ACS data, information that is not easy to find even on the census website. The ACS collects information from about 3 million households per year, compared to the 50 million households that receive the decennial long form. Due to relatively smaller sample size, different lengths of pooled data are required for different regional areas. Major metropolitan areas have annual information in the ACS, while data in smaller areas should be pooled across two, three or five years, making comparisons across regions not as straightforward (see more tips on using ACS in Taeuber 2006).

One other major problem with the census data, the item non-response, is not addressed in detail. Currently, a large portion of census data is imputed and not flagged for the general public. Many people leave some questions unanswered because they fear for their privacy. For example, 20 percent of the wage and salary information was imputed in 2000 (see Hillygus et al. 2006). While the ACS will be able to employ more qualified interviewers due to continuous employment, the item non-response problem will probably persist. The U.S. Census Bureau should consider using flags so that researchers could test for the imputation effects when using ACS data.

Much attention is focused on changing the census race measurement from a forceful selection of one race in 1990 to the freedom to select up to six races at a time. However, this modification creates complexities in measuring the trends over time. A sound example can be found in O'Hare's discussion on the racial composition of U.S. children in chapter 7. O'Hare finds that there is more racial diversity among children now than there was ever before. Unfortunately, an unknown portion of this increased diversity might...