Four (and a Half) Preregistered Failures to Replicate the Weapon Focus Effect in Online Samples

Prior research suggests that witnesses have worse memory for armed compared to unarmed perpetrators, a finding known as the Weapon Focus Effect (WFE). Because many legal professionals and eyewitness experts believe that the presence of a weapon has harmful effects on witness memory, determining the generality of this effect is of great importance. Initially, the current investigation was concerned with identifying moderators of the WFE. However, because we were unable to replicate the WFE, subsequent experiments focused on investigating the effect's replicability in the context of online samples. In total, we conducted five pre-registered experiments ranging from 100 to 800 participants each (total n = 1,316). These experiments used materials from studies which had previously shown significant WFEs and varied in presentation modality (videos or slides), the presence of audio, weapon type (knife or gun), perpetrator gender (male or female), and sample type (student or non-student). Evidence supporting the replicability of the WFE was weak. We did not find a significant WFE for target description or lineup identification in any experiment, and some experiments showed significant anti-WFEs, where participants had better memory for armed compared to unarmed perpetrators. Random effects meta-analyses using data from all experiments found that the effect of weapon presence on both memory outcomes was nonsignificant. Taken together, the current results suggest that the WFE is either not replicable, or is highly dependent on methodological factors such as study modality.