This paper develops an efficient scheme for distributed computation of averages of the node data over wireless sensor networks. The scheme first formulates a data-encoded pulse-coupled oscillator (DE-PCO) model for sensor nodes, in which the node data is encoded into a PCO phase shift, and then it defines a deferred and accumulated coupling (DAC) strategy which describes the data coupling between different nodes. Following the DAC strategy, the coupled DE-PCO converges to a synchronizing equilibrium from which each node can decode the global average of the node data. Compared with the conventional communication network, the DE-PCO-based computation does not require special data routing and medium access control (MAC) protocols. Therefore, the proposed scheme is efficient and easy to implement.