> (That’s how I learned that `log_salary~height*gender`, `log_salary~height+gender` and `log_salary~height:gender` meant completely different things.) Interact

> (That’s how I learned that `log_salary~height*gender`, `log_salary~height+gender` and `log_salary~height:gender` meant completely different things.) Interaction terms ftw :-) I would postulate that every analysis becomes more accurate if you consider the interactions ... but at the same time it makes interpreting (and presenting) the results more complicated.