Nielsen measures TV audiences using a sample of homes installed with People Meters that directly capture not only what programs are being watched, but also who is watching them and when. While this direct measurement is a proven means of estimating TV ratings, target sample sizes can be achieved more efficiently by directly measuring only the programs being watched, then imputing the age and gender of the audience.
To this end, Nielsen developed a measurement approach utilizing a sample of homes installed with a Code Reader meter, which only measures household tuning. Viewing is assigned to the corresponding household members using imputation techniques that incorporate the demographic information collected from the Code Reader households in addition to directly measured person viewing behavior from People Meter households.
As with all samples, Nielsen audience estimates are subject to sampling error. In addition to the usual drivers of sampling error, audience estimates from Code Reader samples include variance due to the viewer imputation process. An analysis was undertaken to dimension the extent to which sampling error is increased by the imputation methodology.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.