With a controlled experiment, the slope can tell you the average change in *y* that would be caused by a change in *x*. With an observational study, however, the slope cannot be relied on to predict the results of interventions. It takes a lot of hard work to draw causal inferences from observational data, with or without regression. pl875