Reversed columns in regression example
On page 173 of the 2nd edition of How to Measure Anything, I show a formula for estimating ratings points of a TV show based on the correlation to weeks spent promoting the show. The formula shown is actually an estimate of promotion weeks based on ratings points instead of an estimate of ratings points based on promotion weeks. The correct formula should be
Ratings points = 0.215*promotion weeks + 0.877.
The subsequent estimate of ratings points from 10 weeks of promotions should then be about 0.215*10+0.877 or 3 ratings points.
Doug Hubbard
