In recent years, well-documented failures in forensic science have led to an increased interest in the development of scientifically valid quantitative analysis methods for forensic analysis. These quantitative methods require clear, concise explanations of statistical concepts that can be understood by individuals with no statistical training. In this study, we assess different explanations of a bullet matching algorithm used to quantify the similarities between two bullets. We conduct a study of potential jurors regarding the use of a bullet matching algorithm and demonstrative evidence in a fictitious trial, where the recovered bullet was the only evidence linking the defendant to the crime. Potential jurors are asked to express their perception of firearm evidence as a whole, and to make informed choices regarding the guilt of the defendant and the reliability of the evidence. This research provides insight into how new methods for forensic assessment can be introduced to jurors. This study also relates to the issues of explaining machine learning techniques, such as random forests, in a way that encourages statistically inexperienced individuals to place trust in the results.