Wireless sensor networks place sensors into an area to collect data and send them back to a base station. Data fusion, which fuses the collected data before they are sent to the base station, is usually implemented over the network. Since the sensor is typically placed in locations accessible to malicious attackers, information assurance of the data fusion process is very important. A witness-based approach has been proposed to validate the fusion data. In this approach, the base station receives the fusion data and "votes" on the data from a randomly chosen sensor node. The vote comes from other sensor nodes, called "witnesses," to verify the correctness of the fusion data. Because the base station obtains the vote through the chosen node, the chosen node could forge the vote if it is compromised. Thus, the witness node must encrypt the vote to prevent this forgery. Compared with the vote, the encryption requires more bits, increasing transmission burden from the chosen node to the base station. The chosen node consumes more power. This work improves the witness-based approach using direct voting mechanism such that the proposed scheme has better performance in terms of assurance, overhead, and delay. The witness node transmits the vote directly to the base station. Forgery is not a problem in this scheme. Moreover, fewer bits are necessary to represent the vote, significantly reducing the power consumption. Performance analysis and simulation results indicate that the proposed approach can achieve a 40 times better overhead than the witness-based approach.