Nonresponse has the potential to induce biases in the estimates, if units that do not respond to the survey are systematically different from units that do respond. The most common way to deal with unit nonresponse is through a weight adjustment procedure. This procedure typically consists of eliminating the nonrespondents from the data file and adjusting the weights of the respondents, with the goal of reducing nonresponse bias. We compare the bias and variance using the current ATUS adjustment to four propensity score adjustment weights: 1) logistic regression, 2) weighting classes based on logistic regression, 3) regression tree using a CHAID growth method, and 4) regression tree using Gini growth method. This paper summarizes a simulation study exploring and comparing alternative response propensity procedures in terms of bias and efficiency using data from the American Time Use Survey.