|Census Bureau Director Kenneth Prewitt revealed the results of the national count on Dec. 28, 2000.
"We've seen just enough to get excited," said Debbie Solomon, senior partner-group research director at WPP Group's MindShare USA, New York. For marketers, the focus will not be on the changing census picture from 1990 to now. Instead, they will be looking at how far off their 1999 estimates-based upon annual Census Bureau population updates and private data analysis-are from actual census data.
Higher ethnic populations
For example, marketers have been anticipating higher ethnic populations and adjusting their plans accordingly. Graciela Eleta, general manager-multicultural business development organization at Procter & Gamble Co., said "Way before the census, or even the prior census, we have been very focused on driving our shares among ethnic consumers. The new census just reaffirmed how important this is for the company."
But there will be new and useful information in the details, as research companies such as Claritas, CACI Marketing Systems and Orange, Calif.-based SRC release their first census-updated numbers. SRC's data were posted on its Web sites (Freedemographics.com and demographicsnow.com) May 11.
Among the first changes will be a re-ranking of
|Read Ad Age's full coverage of census-related marketing issues.
Media ratings reports
By September, it will begin projecting TV ratings based on the new numbers and the Audit Bureau of Circulations will begin to issue audit reports on magazines and newspapers using the new numbers. Also in September, Arbitron will switch to numbers based on the 2000 census, though its first radio audience reports based on the new data won't be published until January.
While the Census Bureau isn't scheduled to begin releasing detailed household, income, language and age breakdowns that marketers prize until later this year or in some cases next year, significant differences are already surfacing.
The biggest is that the nation's total population is larger than expected. Both the Census Bureau and private data analysts that base their models on the bureau's numbers underestimated the U.S. population by about 2%. The Census Bureau said there were 281,421,906 people in the U.S. on April 1, 2000.
Some of the findings
Most intriguing for marketers, the underestimate is not consistent.
- By age: The number of 18-to-29-year-olds counted by the Census Bureau on April 1, 2000, was 4.85% higher than its estimate for November 2000 (issued in January 2001 based on projections from the 1990 census). But the bureau was on target in predicting the over-50 population. In two other age brackets, under-18 and 30-to-49, the bureau's previous population projections were about 2% below the 2000 census tally, the same margin of error as in its total population estimate.
- By metropolitan regions: There was a great deal of difference between census results and previous estimates from city to city, and some indication that, while California's big-city growth had been correctly predicted, growth in some other big cities had been badly underestimated. Census data from 2000, for example, show the Las Vegas area had 13.2% more people than were projected on July 1, 1999; Austin had 9% more; and New York City; Dallas-Fort Worth, Texas; Phoenix; Denver; Tampa-St. Petersburg and Orlando, Fla.; Charlotte, N.C.; Greensboro, N.C.; Providence, R.I.; Raleigh, N.C.; and West Palm Beach-Boca Raton, Fla., all had at least 5% more people than had been forecast. One major city, San Diego, was smaller than estimates, though barely.
- By ethnic groups: The U.S. Hispanic population, as widely reported, was dramatically underestimated, though not quite as underestimated as some accounts comparing 2000 numbers with 1990 might suggest. The Census Bureau found 35,305,818 Hispanics on April 1, 2000, a 9.9% increase from the 32,838,000 that the bureau had estimated in November 2000 based on projections from the 1990 census. In a new report on its 2000 data, the Census Bureau reported last week that half of all Hispanics lived in California and Texas (and that 76.8% lived in those two states plus Florida, Illinois, Arizona and New Jersey). It also said the Hispanic population in 2000 was on average nearly 10 years younger than the overall U.S. population.
- By race: The data on race in the 2000 census aren't exactly comparable to the 1990 census. The Census Bureau's decision to allow multiracial choices for the first time in the 2000 census means that marketers will have their first numbers on people of multiple races, but they will need to do some number crunching to do any look-back comparisons. Further comparisons may not be easy. The Census Bureau, which had estimated 33,619,000 million African-Americans in America on Nov. 1, 2000 (based on 1990's census), reported its April 1, 2000, count was 34,658,626, a 3.1% increase. However, the bureau also reported 6,826,226 people listing two or more races.
The race revision definition may also make comparisons of the Asian population difficult. The Census Bureau, which had estimated 11,279,000 Asian and Pacific islanders were in the U.S. on Nov. 1, 2000, based on 1990 numbers, found 10,641,838, a 5.6% decrease. However the difference could be due to people not counted in that number who listed multiple races.
The intrigue may not be over. Census Bureau officials are trying to determine whether the differences between previous projections and the 2000 census data reflect better counting in 2000, undercounting in 1990, or underestimates of immigration. Census officials said the Hispanic numbers, in particular, show growth, but also unexpected changes in where Hispanics live that weren't reflected in their earlier estimates.
Signe Wetrogan, the Census Bureau's assistant chief of population estimates, said that increased size of the overall population and the greater than expected Hispanic number were the big surprises of the 2000 count. Not surprising was population growth in the U.S. South and West that the data portray.
The Hispanic 'puzzle'
"Hispanic, that is our puzzle," she said. "We have started to see growth of Hispanic populations in non-traditional counties," noting, as an example, growth in North Carolina, not traditionally a heavily Hispanic area.
The Census Bureau isn't scheduled to release details on Hispanic households that speak Spanish at home until next year, but it may provide some early data in an effort to understand why the overall Hispanic population total is so different than had been expected. Some of the differences, Ms. Wetrogan said, are due to "differences in coverage," but the bureau is trying to understand exactly what happened.
"Some components of changes, like births and deaths, are well measured and well estimated," she said. "Some, like international migration, are maybe less well measured. Internal migration is even less well measured."
Demographic research companies
Most marketers don't look at census data but instead use models or analysis supplied by major demographic research companies. The research companies project information on target market areas using census data but combine the census numbers with data from other sources. But over the next year, the companies that do those projections have a problem.
Enough census statistics are out on population size and ethnic and race distribution to warrant changes in the information going to clients. However, much of the census information thus far released is limited to the voting age population. Among the data not out is information on the numbers of households and on household income, nor has much information been released on age breakdowns. That leaves demographic companies having to project the rest of the changes based on 1990 numbers and then having to correct any errors when actual census statistics come out.
Fred Wilcox, VP-demographics resources at San Diego-based Claritas, which produces data through its Marketing Statistics and PRIZM units, said "The possibilities are there was an undercount in 1990, an overcount in 2000, or potentially immigration was far higher than what we thought."
How significantly local market numbers will change this year, then, depends on the demographic models used.
"For the most part, the decennial census doesn't give us huge surprises at the national level," said Ken Hodges, Claritas' director of demography, noting that the Census Bureau's annual demographic supplements usually give a pretty good idea of what is happening nationally. "It was clear there was going to be an increase in Hispanics. The magnitude of the increase is the surprise. It was notably higher than what the Census Bureau was expecting."
In looking at its own data, Mr. Hodges said Claritas found several differences besides the overall growth in population and the growth in the Hispanic population. The District of Columbia population is larger than was expected, Mr. Hodges said, and Claritas will deal with the change in race numbers by calculating a comparison, while shifting soon to the new racial breakdowns from the 2000 census.
Explaing data gaps
At Arlington, Va.-based CACI, Director of Data Development Lynn Wombold said the company will report not only what numbers changed but how much of the change is due to the Census Bureau's figures being off vs. how much is due to CACI projections being wrong.
The new census numbers also require rebalancing of various household panels used by market research and ratings firms to track consumer behavior. Those changes will begin this year, but some of the resulting numbers won't be reported until next year or later, say market research companies.
Julian Baim, chief research officer of Mediamark Research, said because his company reports survey results in May and November, it wouldn't begin to select a sample based on the 2000 results until October and the resulting changes would not appear until next year.
Readjusting Hispanic numbers
Nick Sorillo, senior VP of AC Nielsen's Homescan product, said his company has made changes to reflect the increased Hispanic population and would be making other changes to reflect the census in its new numbers. "As soon as the census numbers came out, we readjusted the Hispanic numbers," he said.
On the media side, Paul Donato, chief research officer at Nielsen Media Research, said Nielsen's examination of the census so far has shown some trends. "Overall, what we see is about a 1.7% higher population across the board, except in California [where the growth was as anticipated]," he said.
Arbitron Co. said it will implement changes in market size, racial and ethnic breakdowns during its fall survey of radio, whose results are reported in January. Next year, it will introduce changes based on age and sex, which will be used for its winter audience survey. The new census "is not going to change anything because we already do so many things to make sure we're representing Hispanics and other racial and ethnic groups," said Ed Cohen, VP-domestic radio research.
If the speed of the rollout of the new information is clear, the effect the new numbers will have on marketing is less so. Will marketers quickly rearrange advertising to reflect the altered rankings of media markets and ethnic breakdowns, or has much of that already happened?
"It's pretty clear that it is a huge shift taking place," said Randy Melville, VP-urban marketing at PepsiCo's Pepsi-Cola Co., citing the growth, especially of Hispanics and other minorities. "You don't even need a lot of interpretation [to see the shift's impact]. It's like one and one is two." Already, Pepsi and Coca-Cola Co. have launched inner-city initiatives designed to lure young people before they develop brand loyalties, as well as to engage those who feel mainstream American marketers ignore them. "Speed equals share," he said.
Kristen Simmons, VP-marketing at Mazda North American Operations, said the carmaker already was increasing its activities. "In general, we have been upping our diversity efforts over the last couple of years," she said. Mazda plans to dial up those efforts for its 2002 ad budget. "It's not just because of the census shift, but more importantly, as we worked out our psychographic target, people who have that psychographic have skewed to Hispanics and Asians."
Shaking up the tree
P&G's Ms. Eleta said that although P&G has anticipated much of the findings, "I do think for other companies this is going to shake up the tree, and I do expect a continuation of the Hispanic media growth that we've been seeing in the last few years. I think other companies that have not yet joined the fray will. For us, I don't think we're really changing anything as a result of it."
Hispanic media officials are hopeful the census will bring changes.
"I believe that if it is a 10-step process, [the census information] jolted people into thinking about it," said Tom McGarrity, co-president of network sales for Univision. "For people who are on the fence, it got a lot of wheels turning."
Univision isn't waiting for the wheels to turn. It sent out a tape of TV news coverage of the census report on Hispanic numbers to 500 marketer CEOs. "Our pitch isn't that we want them to add Hispanic. It's that we want them to redistribute their money to get 10% of the U.S. market."
~ ~ ~
Jack Neff, Hillary Chura, Jean Halliday and Cara Beardi contributed to this report.