We present a two-tier composite model of node mobility that captures group behavior in a mobile ad hoc network. The first tier represents individual node movement and is based on an autoregressive model of mobility. The second tier captures group mobility behavior by considering correlation among node mobility states. Based on the two-tier model, we propose a scheme to detect the presence of groups among the nodes of a network by performing a correlation index test on the node mobility states. We also propose a group-based mobility estimation scheme, which uses the mobility state of a representative node in a group to estimate the mobility states of the rest of the group members. The group estimation scheme can significantly reduce the amount of data collection required to track nodes exhibiting group mobility in mobile ad hoc networks. The two-tier mobility model and the group detection and estimation schemes are validated through simulations and with GPS data.