|
Activity Number:
|
245
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #306743 |
|
Title:
|
Statistical Modeling: Science versus Business and Domain Expertise
|
|
Author(s):
|
Igor Mandel*+ and David Hauser
|
|
Companies:
|
Media Planning Group and Media Planning Group
|
|
Address:
|
9-15 Berdan Ave., Fair Lawn, NJ, 07410,
|
|
Keywords:
|
statistical modeling ; advertising ; ROI ; optimization ; budgeting ; yield analysis
|
|
Abstract:
|
The dichotomy between results derived from statistical modeling and business expectation is addressed: how to balance unconstrained statistical models that may contradict domain knowledge or expectation. If differences exist, the results need to be defended by the statistician to change that domain knowledge/expectation, or the application of constraints is needed to "correct" the results. This process of alignment is generally informal and usually quite subjective. All statisticians face this dilemma almost every time they present results to an audience. The combinations of "subjective" and "objective", "derived" and "desirable", "free" and "forced" aspects of the modeling could be complicated, and often pose difficult scientific and moral challenges. Different approaches are considered; recommendations follow from extensive advertising ROI modeling created for many product categories.
|
