Nielsen began weighting its national TV ratings sample effective with the 2003-04 season to adjust for significant inconsistencies between its sample and the U.S. population. At the start of the 2004-05 season, it introduced weights to adjust for the under-representation of Spanish-speaking households in its sample.
While controversial, weighting is seen as a pragmatic - and to some, a necessary - solution to making Nielsen's sample more representative of the U.S. population it is designed to measure TV viewing for.
advertisement
advertisement
When a segment of Nielsen's sample - such as Hispanics, of households with children - are known to be out of balance with the U.S. population, Nielsen utilizes mathematical methods that give those sample households the equivalent "weight" of their counterparts in the U.S. population.
While weighting is not necessarily a bad solution to a problem, some Nielsen clients believe the practice shifts Nielsen's focus away from working to make its sample truly representative.
"Moreover," says one client, "to my knowledge, Nielsen has yet to produce a single piece of objective evidence (i.e., independent methodological research such as a coincidental) to demonstrate that they have reduced or avoided sample bias through their [weighting] procedures."
The problem is especially vexing for a category like digital cable, says one observer, because unlike Hispanics, or family size Nielsen has no independent measure such as the U.S. Census to base its weights on.