so far, we have assumed that each predictor has an independent association with the mean of the outcome now we will look at conditioning this estimate on another predictor using interactions fitting a model with interactions is easy, but understanding them can be harder 7.