Which research method assesses how well one variable predicts another without demonstrating a cause-effect relationship between the variables?