In traditional wireless sensor networks, there are many issues potentially affecting their operation and worsening the quality of service in real-world applications. One of the most serious is the decreased precision of sensor data possibly caused when sensors detect and respond to inputs from a highly noisy environment. Thus, the measured and the real value of the observed physical quantity may significantly vary from each other, which may cause serious problems in many real-world applications. In this paper, we consider scenarios where sensor readings are skewed due to Gaussian noise with a variety of standard deviations. More specifically, we apply the generalized Metropolis-Hastings algorithm to suppress this inaccuracy in wireless sensor networks and thus analyze how this algorithm can compensate for incorrect sensor data in various scenarios. This means that the execution of this algorithm is bounded by a stopping criterion designed for wireless sensor networks, and we vary the initial configuration of both the algorithm and the used stopping criterion in order to examine whether the Metropolis-Hastings algorithm can be used to suppress incorrect sensor readings (by comparing with a scenario when no data aggregation is applied) and identify the optimal initial configuration of the algorithm and the applied stopping criterion. Also, the performance of the algorithm is compared to the Best Constant weights, the algorithm lately identified as the best-performing consensus-based algorithm with the Perron matrix for this purpose.