[kryogen1c]

it really is telling that articles like this arent laughed off the face of the planet.

if this were a univariate, observational study of a medical condition claiming "some correlation" with an r2 of .3, the authors would be fighting mobs with pitchforks and torches.

its such a cheap play at current politics. if people cannot discern this from science, we truly are lost.

[AstralStorm]

Weak correlations are fine, as long as they're consistent and statistically significant.

This generally means that there is a factor causing a misfit with the simplistic base model. It's fine if the factor is properly identified, which it is not in this case. I think a logarithmic fit would give much higher R2 than linear. (Logistic fit is commonly used for prices and salaries.)

This problem goes away with multifactor analysis as you see correlation in errors or not, or using nonlinear least squares, NLS.