data collection: Passing the Torch

By Published on .

Census 2000 arrives this month, with some departures from tradition.

The first Census form of the millennium is on its way to a mailbox near you, but relax - five out of six Americans will have to answer a measly seven questions (six related to population, one on housing) about themselves and their households. Even the rest who get the 53-question long form probably won't need to use a lifeline to plow through it. In fact, the U.S. Census Bureau has spent more than five years designing a form that's user-friendly. A new font, crisp colors, and informative icons are just some of the changes that make this year's form different from the one in 1990, which some called "an SAT approach to the Census," says Carolyn Hay, program analyst for the U.S. Census Bureau. "We're not blaming the design of the 1990 form for the low response rate. But we wanted to do everything we could to maximize response in 2000."

There are more than just cosmetic changes to Census 2000, however. Scan the questions and you'll notice several revisions since 1990. To date, the media, including this magazine, has focused on probably the biggest change: For the first time in Census history, Americans will be able to identify themselves as more than one race (see "A Close Race," January). But there are other important shifts as well - a subtle rewording here, a new question there. The following highlights the most significant differences between the 1990 and 2000 Censuses - changes that will provide deeper insight into the nation's population for decades to come.

Mixing Things Up: Questions 5 and 6

Since 1970, respondents have identified their race on the census form before giving their ethnic origin (Spanish, Hispanic, or Latino). A major problem arose with this system in the 1990 Census: 10 percent of Americans failed to filled in the question on ethnicity. What happened? When many Hispanics answered the question on race, they wrote in responses such as "Hispanic," "Mexican," and "Puerto Rican" - unacceptable answers to the Census Bureau because it classifies such identifications as ethnic origins. When those same respondents then encountered the question on ethnicity, they were often confused and left it blank, says Jorge del Pinal, assistant division chief for special population statistics at the U.S. Census. "It was suggested that if Hispanics were given the ethnic origin question first, they would have less trouble answering the race question later," he says. The bureau took up the suggestion and moved the question on ethnic origin ahead of the one on race for this! year's count. Directions clearl y instruct people to answer both.

The change proved successful at the 1998 dress rehearsal sites in South Carolina; Sacramento, California; and Menominee, Wisconsin. Of the three sites, Sacramento scored the highest non-response rate, with 5 percent of the population not answering the question on ethnic origin. Menominee had the lowest, with 0.6 percent.

Lower non-response rates for ethnic origin may come with a price: higher non-response rates to the race question, especially among Hispanics. At the dress rehearsal site in Sacramento, nearly 22 percent of Hispanics failed to answer the race question, compared to the non-Hispanic non-response rate of 0.5 percent. Del Pinal contends the tradeoff is a good choice. "It's unfortunate that we'll have to impute the race for many Hispanics, but [at the dress rehearsal sites] we got much higher response rates for ethnic origin," he says. "I don't think there will be much crying about this at all."

Yours, Mine, and Ours: Question 2

There's no denying that few families today resemble Ward and June Cleaver's on Leave it to Beaver. The Census Bureau acknowledges it, too. In the 1990 Census, residents of a household picked one out of 11 choices to best describe their relationship to the householder. Those selections included husband/wife, brother/sister, and father/mother. This year, there will be four additional options. Two of the new choices reflect an expansion of the non-nuclear family: "parent-in-law" and "son-in-law/daughter-in-law." Neither choice appeared on the 1990 form.

The other new selections are actually spin-offs. "Foster child," which was previously lumped in "roomer, boarder, or foster child," is now on its own. This distinction will allow the government to get an accurate count of the foster-child population and the households caring for them. The Census Bureau has also separated the "natural-born or adopted son/daughter" category into "natural-born son/daughter" and "adopted son/daughter."

Hallmark Cards Inc. wishes the Census Bureau had made these changes ten years ago. The company recently launched a line of greetings cards called "Ties that Bind" that focuses on the growing population of stepfamilies and other nontraditional households. "Typically there is reliable Census information for each of our product markets and we certainly use those figures," says Jay Dittmann, vice president of research at Hallmark. But, he adds, government data on stepfamilies and children has been less than exceptional until now.

When Mom's Mom Plays Mom: Question 19

We all learned when we were little that if mom says no, you go ask grandma. And while it may have taken the government decades to figure this out, it will finally ask about grandparents in 2000. Specifically, the Census Bureau wants to know about grandparents as caregivers. A new three-part question asks first if any child under age 18 in the household is living with his or her grandparent. If so, the respondent must answer whether that grandparent is responsible for most of the basic needs of the child and, if yes, for how long.

Kraft Foods, Inc., a company that targets its marketing efforts on female heads of household, will be keeping a close eye on the results of question 19. "Understanding the distribution of grandparents as caregivers will infuse our thinking about how to approach this market creatively," says Mary Carroll, director of consumer insights and strategy at Kraft. "We will follow the information very closely when it comes out."

Disability Examiner: Questions 16 and 17

Today, at least one in five Americans reports some level of disability and soon we'll know a lot more about them. In 1990, the Census inquired only about the disability status of people 15 years and over. Census 2000 lowers that bar by ten years and will collect data on the disability status of children aged 5 and over, as well as adults. This revision shifts the focus away from the impact that disabilities have on work and looks more at the effects disabilities have on major life activities such as seeing, hearing, and the ability to perform physical and mental tasks.

While two other governmental sources have information on persons with disabilities - the Survey of Income and Program Participation (SIPP) and the National Center for Health Statistics (NCHS) - neither meets the demand for data that only the Census Bureau can provide. "There's certainly a need to collect disability data in the decennial census because people at the state and local level want to know about people [in their community] with disabilities," says Jack McNeil, statistician at the U.S. Census Bureau. "SIPP and the NCHS collect data only at the national level."

In this article:
Most Popular