Asked in Math and ArithmeticStatistics
When you use your regression line to predict values that are not in the sample and discovered large residual values in each prediction what can you say about your regression line?
July 27, 2012 11:12AM
There are several possible explanations: Leaving aside the two most obvious reasons: calculation error and attempted extrapolation, there are the following possibilities:
The true relationship is non-linear.
A relevant variable has been missed omitted.
The observations are very variable: leading to a very large residual error.
There is not enough variation in the independent (or predictive) variable so that Sxx is very small.