Navigation: Media Wisdom :

Estimation vs. Tabulation

Contents

TView predicts the delivery of a proposed television plan by estimating who will see our campaign how often. This is quite different than looking at Nielsen records and tabulating who watched what in the past.

	Tabulation of Past Schedule	Prediction of Future Schedule
Direction	Look to past	Predict to future
Useful for...	Post-buys and stewardship	Devising future media plans, predicting their results, and refining them
What does that mean?	Past viewing is tabulated, that is, directly counted: we tally each distinct time that one of our spots was reported as having been viewed	Pattern of past viewing is used to estimate each respondent’s probabilities of commercials in a future viewing
Core elements evaluated	Individual specific spots in specific programs	Description of proposed plan by dayparts, networks, and other plan entities
Example systems	NPower, Ad*Views	TView
When it's desired	Takes into account quirks* of viewing on specific air dates to create a good estimate for an exact, specific, past schedule	Unaffected by quirks* that occur at one point in time so as to produce good estimate of a future schedule.
When it's desired	* Quirks: Real-life events (e.g., news events, holidays, specials, schedule delays and pre-empts) which can cause a specific schedule on specific dates to have an unusually light or heavy reach, compared to what would be the best estimate of what to expect on average.
Statistical “replacement”	Tabs viewership of specific spots. You might compare this to statistical draws without replacement.	Equivalent to statistical draw with replacement. Continues until entry GRPs have been satisfied.
Huh?	We know that there are 5 white cars and 5 red. We observe 4 red cars and 5 white cars as they pass. We know for certain that the next car, the last, will be red. We can say nothing at all about the 11th car that passes.	5 white cars and 5 red cars have been observed. The odds that the next car (the 11th) will be red are 50%. The same is true for the 12th, 13th, and so on.
What happens as number of spots approaches the number of spots available?	Tabulation proceeds spot by spot, and ends when last spot is counted.	Original number of spots reported in the source data sets the possible cume max, defined as persons who have a non-zero chance of future exposure. Build-up proceeds using personal probability against respondents.