Per Arbitron's promise to subscribers last year, the minimum sample size in each market (a certain proportion of the target sample size) increases over time, with targets rising incrementally depending on how long PPM has been in service. In April, Arbitron failed to meet scheduled goals in Los Angeles, Chicago, Detroit, and Long Island (Nassau-Suffolk County). In Los Angeles, the 18-34 African-American sample size was 64% of the target, versus a minimum goal of 75%. In Detroit the sample size for the same demo was 59% of the target, when Arbitron hoped to achieve at least 70%. In Nassau-Suffolk, the 18-34 Hispanic sample size was 57% of the target, versus a minimum of 80%.

Arbitron has made a number of moves to increase sample sizes for minority audiences, including bigger incentives for panelists, recruiting more panelists from cell phone-only households (to achieve higher penetration among young people) and replacing non-participating panelists.

Nonetheless, it can ill afford more shortfalls in minority audience measurement. After months of criticism from minority broadcasters, who claim that underrepresentation of these demos is resulting in steep drops in their apparent audience sizes, earlier this month the Federal Communications Commission opened an inquiry into Arbitron's PPM ratings. Democratic commissioners ominously hinted that the FCC may be empowered to take regulatory action against the company -- perhaps to the extent of enforcing an indefinite embargo on the use of PPM ratings.

This isn't Arbitron's first run-in with the powers that be. Last year the same complaints prompted the attorneys general of New York, New Jersey, and Maryland to bring lawsuits against Arbitron, alleging that the flawed sample methodology constituted a violation of various civil rights statutes. These lawsuits were all eventually settled out of court, with Arbitron promising to improve sample accuracy.

To round out the bad news, last week Arbitron revealed that it will delay the release of ratings from New York City for April because an incorrect population estimate was manually entered during the ratings calculations. Furthermore, similar errors apparently occurred in ratings for October 2008 and for the period from December 2008-March 2009. New figures for the last period will be issued on June 8.

This news story is important. I recommend that everyone read it.