Quote:
Originally Posted by cannabusto
Also, I still feel there's no rebuttal to the fact that these methods are useful in soft sciences despite some assumptions not being met and the fact that things are much more difficult to measure than in hard sciences.
I do believe people trained in hard sciences have better quant backgrounds in general. But I also know that grad programs in the soft sciences teach many advanced analytics techniques like factor analysis/PCA, many flavors of regression and classification, network analysis, etc, etc. And they're often quite good at dealing with messy data which is far more common in the soft sciences.
These are very basic techniques, nothing advanced about them. Blindly applying them leads to conclusions worse than those given by common sense or dumb luck.
Almost no one in the "soft sciences" is as good as your typical engineer or scientist at data analysis. Not even remotely close.