Nielsen Admits Discrepancy in Viewer Surveys : Ratings: The media monitoring firm concedes a probable undercount. Spanish-language channels cite loss of ad revenues.

An executive of the TV ratings service Nielsen Media Research acknowledged Wednesday that its audience figures for the Los Angeles area probably undercount viewers of Spanish-language stations, but Nielsen said it has no plan to alter its method of collecting data.

The Nielsen measurements have a big impact on television station revenues because sponsors make advertising decisions based on audience size.

At issue is the discrepancy in the results of two surveys Nielsen conducts to measure television audiences in the market covering Los Angeles, Orange, Ventura, Riverside and San Bernardino counties. One is a conventional general-market study of 475 households meant to represent a cross-section of the population. The other, begun several years ago at the request of Spanish-language broadcasters, concentrates of the viewing habits in Latino households.

During February, for example, Nielsen’s general market sampling found that 196,000 households were tuned in nightly to the 6 p.m. news on Spanish-language KMEX-TV (Channel 34). According to the other survey, more than 300,000 Latino households were watching. The difference was 54%.

The lower numbers have been costing KMEX a “huge” amount of advertising revenue, station manager Tom Arnost told news reporters. Television stations in the local market bring in more than $1 billion in advertising revenue each year, Arnost said, but less than 10% of that business is going to KMEX and its Spanish-language rivals, KVEA-TV (Channel 52) and KWHY-TV (Channel 22).

Appearing with Arnost and other KMEX officials was Ceril Shagrin, a Nielsen senior vice president. He said the Latino survey remains “the best estimate currently available for measuring Hispanic viewing” and is better than Nielsen’s survey of the overall market.

But Shagrin stopped short of calling the whole-market data incorrect. She said Nielsen is committed to improving the system it uses to gather that information, but would not give details.

Any adjustment of Nielsen’s method of determining ratings here would subject all of its multimillion-dollar contracts to renegotiation. The English-language channels would be likely to resist any change that would deflate their viewership, Arnost said.

Nielsen measures television viewing in 220 cities across the country, using random-sampling methodology in most of them. The company also uses the Latino survey in 10 other cities, but Shagrin said it has not seen similar discrepancies elsewhere.

The difference apparently arises because the survey covers more Latino households than the general-market survey--200 compared to 115--and takes greater care to reflect Census Bureau figures for the proportion of households in which Spanish is the first language. The general-market survey is more heavily weighted with English-speaking viewers than the population as a whole.

Maureen Schultz, research director at KMEX, said that if the results of the Latino ratings were substituted for the Latino portion of the general-market survey, KMEX’s 7-to-11 p.m. ratings for February would have been 70% higher, while every English-language station would have been down as much as 10%.

Schultz said that some discrepancy between samples is to be expected but that the Nielsen results are too far apart.