An alternate hypothesis is that while students might be performing on the ITS across several days, they are not forgetting the previously learned material. Rather, the students are just making a mistake on the first question of the day (rustiness effect) after which they no longer slip at a higher than usual rate. So the low accuracy on first attempt on a new day might not be captured in the forget parameter, it could be that they just slipped and answered wrong. This explanation makes it quite necessary to look into the slip parameter. The KT-Slip model is similar to the KT-Forget model and can be represented simply by connecting the time node to the slip node instead of connecting to the forget node as in the Forget model. The Slip model allows us to model the different slip rates of the new days and the same days. Since the slip node is only conditioned on the added new time node, there is also one new parameter slip_n introduced in this KT-slip model, which represents the slip rate on a new day, and the original slip parameter is denoted as slip_s here. When a new day response occurs, New Day=T, the probability of slipping is slip_n, P(slip=T|New Day=T) and is slip_s, otherwise.