Nielsen: We Undercounted People Of Color, Not Blacks, Hispanics, But Asians

A report released late Thursday by Nielsen Media Research reveals just how complex it can be to accurately measure multicultural audiences, even within a single household. It also reveals that Nielsen's local people meter sample in New York did underrepresent a major ethnic group. It just didn't happen to be either Hispanics or African Americans, but Asian Americans.

The report, an unusual public disclosure by Nielsen on the nature of its processes and procedures, summarizes the findings of the Media Rating Council's audit of the New York people meter system and sample. Nielsen said it released the report in response to issues raised by the MRC, its audit, and the subsequent leak of that report earlier this week to the Los Angeles Times.

Ironically, the findings do not support the contentions of News Corp.-backed opposition group Don't Count Us Out--which has maintained in a series of TV and print ads, as well as public statements and aggressive lobbying efforts, that the MRC audit proved that Nielsen undercounts people of color, that the people meter technology is flawed, and that the local people meter service should be delayed.

Actually, Nielsen said the report did reveal an underrepresentation of some people of color in its New York people meter sample: Asian Americans, who were 3 percent below the market's universe estimates for the demographic group. Ironically, no Asian-American groups have voiced any opposition to the local people meter service. The service went live on June 3, and Nielsen has faced a groundswell of pleas from clients--mainly major broadcast groups--and trade associations to delay the service until the issues can be resolved and the MRC grants accreditation.

Nielsen executives said it is continuing to work with the MRC toward that end, and expects and hopes to have a breakthrough in New York and possibly even Los Angeles soon. But it seems unlikely that Nielsen will receive MRC accreditation in Los Angeles prior to the service's July 3 launch date-- although that service may be delayed in any case, depending on the outcome of a July 1 Superior Court hearing to grant an injunction that would delay the service. That injunction was the result of a lawsuit filed by Univision Communications, which claims the rollout of the Los Angeles service will damage its standing in the ad community.

Meanwhile, the report released by Nielsen on Thursday says that only four of 85 categories audited by Ernst & Young for the MRC were "tagged," including:


* Cable TV information (15 of 602 homes audited were affected by cable rebuilds)
Race identification (discrepancy in 2 homes out of 30 audited)
Ethnic identification (discrepancy in 2 of 27 homes audited for ethnicity)
Incomplete documentation (applies to more complete disclosure in Nielsen's standard reports).

Nielsen said the MRC also expressed concern about higher-than-normal fault rates, and that the MRC has formed a special "faulting committee" to work with Nielsen and come up with solutions to the problem. Faults refer to situations in which the people meters fail to report data for a variety of reasons--including equipment problems, communications errors, unplugged meters, or improper usage of the meters.

The MRC audit did find that the fault rates were slightly higher for African-American households (25 percent) than for the overall population (16 percent), and Nielsen said it is looking into possible causes, implying that one might ironically have been DCUO's advertising and public information campaign targeted at African-American households in the market.

"Starting in late March, the News Corp.-backed advocacy groups took a strong public position against the use of People Meters," asserted Nielsen. "Broad-based mailings, telephone messaging, television coverage, and possibly door-to-door canvassing may have led respondents to become less cooperative. We are in the process of assessing the impact of the controversy on our sample quality."

In its disclosure, Nielsen did acknowledge some discrepancies with the MRC's audit in identifying the race and/or ethnicity of some sample households, but said it only goes to prove "how complex it can be to gather accurate data from multiracial, multiethnic, or Black-Hispanic households."

Asked what her household race or ethnicity was, one respondent told Nielsen's field representatives "I don't know," citing the fact that her parents were "white and black and her grandfather was Cuban." The respondent ultimately identified herself as African American, and changed the household's designation from Hispanic to non-Hispanic.

In another case, the MRC audit found a household to be Brazilian and the language used to be "only English," since both members of that household spoke mostly Portuguese and English, but did not speak Spanish. Nielsen's records showed the household to be classified as Argentine, speaking mostly Spanish.