To choose an outcome, judgment aggregation methods often rely on the number of votes each formula receives from agents. This implicitly assumes that all agents possess equal reliability and that all votes carry identical weight. In this work we consider an epistemic view of judgment aggregation, where we consider that there is an underlying truth. Finding the truth by using what the majority of agents says can lead to a wrong solution, i.e. a solution where some of the formulae do not have the correct truth value. The idea of this work is to follow the opinions of the most reliable agents to find the truth. To this aim, we propose a new family of judgment aggregation methods that evaluate the reliability of the agents and issues. This evaluation is then used to take a decision and find the truth, instead of simply considering the number of votes. We provide an experimental study showing that these methods yield superior results in the truth-tracking task compared to existing approaches in the literature.