• by Joe Mandese  @mp_joemandese, May 2, 2007

In yet another surprise announcement, Nielsen Media Research Tuesday unveiled a plan to further weight its national people meter ratings sample in an effort to make it more representative of the national population. The decision, which appears to have been made without consulting its clients, is bound to spark a debate over the validity of the decision and its impact on the national TV ratings Nielsen produces.

While weighting - a mathematical adjustment that assigns greater value to the audience data provided by underrepresented types of respondents - isn't necessarily deemed bad research practice (in fact, many consider it a necessary evil), it can impact the reliability of ratings results. Just such a phenomenon occurred when Nielsen introduced time-shifted viewing data in January, and some bizarre anomalies were discovered, such as lower audience totals for ratings that included multiple days of playback data than those reported by "live"-only ratings. Those glitches were attributed to vagaries generated by Nielsen's existing weighting schemes.

In Tuesday's announcement, Nielsen said it will change its national weighting model to include "Black/Non-Black and Hispanic/Non-Hispanic household controls, as well as persons' age and gender controls within Hispanic households," effective in September 2008. The decision likely reflects the fact that Nielsen is having trouble recruiting and/or retaining enough of such households to participate in its sample.

Nielsen also said its is evaluating whether to begin weighting results for households equipped with certain digital media, including digital video recorders and digital cable television services. That decision, Nielsen said would be made by August following an evaluation by its statisticians, and that clients would have an opportunity to evaluate the impact of those changes with a year's worth of evaluation data before they are implemented as part of its official ratings in September 2008.

That evaluation indicates Nielsen may also be having trouble recruiting a sufficient number of households equipped with state-of-the-art digital media technologies such as digital set-top devices to make its sample representative with the U.S. population.

Last week, Nielsen published it first official estimate for U.S. DVR penetration - 17.2% -- which is somewhat higher than the 15.8% of Nielsen's national TV ratings sample that are equipped with DVRs.