Linear regression r square

r square is a measure of the linear relationship between the variables. The nearer r squared is to one, the stronger the linear relationship. However, a linear relatinship is NOT a causal relationship. Also the absence of a large r square is not evidence of no relationship: there may well be a non-linear relationship.

It is commonly accepted that for scientific purposes, a very minimum of 0.98 is accepted for r2. This is because 0.98 actually allows for quite a variance (by eye)