In this paper, we study the implication of wireless broadcast for data aggregation in lossy wireless sensor networks. Each sensor node generates information by sensing its physical environment and transmits the data to a special node called the sink, via multihop communications. The goal of the network system is to compute a function at the sink from the information gathered by spatially distributed sensor nodes. In the course of collecting information, in-network computation at intermediate forwarding nodes can substantially increase network efficiency by reducing the number of transmissions. On the other hand, it also increases the amount of the information contained in a single packet and makes the system vulnerable to packet loss. Instead of retransmitting lost packets, which incurs additional delay, we develop a wireless system architecture that exploits the diversity of the wireless medium for reliable operations. To elaborate, we show that for a class of aggregation functions, wireless broadcasting is an effective strategy to improve delay performance while satisfying reliability constraint. We provide scaling law results on the performance improvement of our solution over unicast architecture with retransmissions. Interestingly, the improvement depends on the transmission range as well as the reliability constraint.