[Home Page] [Previous Page]
[Old Roundtable Archives] [New Roundtable]
Posted by Lisa on January 10, 2000 at 19:12:02:
In Reply to: Re: Negative Beta Values posted by Kent on January 10, 2000 at 18:26:10:
Thanks. That was my basic understanding, but let me give you a specific example where it seems to go against logic, but which has come up in a number of our studies.
Assume you're looking at customer satisfaction with cars. Using a multiple regression analysis (R2=0.87) and beta values as such:
Independent Attributes: Satisfaction with...
color of car 0.457
gas mileage 0.654
comfort of seat -0.245
Dependent Attribute: Overall sat with car
Do I really then interpret that the more likely their seat is to be comfortable the less likely they are to be satisfied with their car?
Or, does a negative beta value in this instance simply imply that the independent attribute is so inconsequential that the effect shouldn't be factored into the results, or is it a flaw in the research design/questions, or something else entirely?
Thanks!
Subject: