What is a lack of fit sum of squares analysis?

1 Answer
Feb 19, 2018

Basically, just the inverse of the correlation coefficient.


Normally, we are interested in how well the points fit a modeled line. In a sum of the squares analysis that is the correlation or regression coefficient.

If you want to see how poorly the data fit a model, you could take the inverse. Then, a "high" number (anything over 1.2 for example) indicates a POOR fit.

Personally, I don't see any value in the practice.

For example if #R = 0.3547#, which would be a poor correlation, the inverse is #1/R = 2.82#