Memory for a witnessed event can be reshaped by information encountered afterwards, a phenomenon known as the misinformation effect. Whether this distortion depends on which details are targeted or which individuals are tested remains unclear. The present study was a planned re-analysis of the trauma-film experiment of Sharma et al. (2024), in which 243 adults watched a distressing film, heard an audio summary that misrepresented, repeated, or omitted 18 critical details, and completed recognition and source-memory tests immediately and one week later. Using generalized linear mixed models (GLMMs) with crossed random effects, the study tested whether the misinformation effect was moderated by detail centrality (central vs. peripheral), image-based visual salience estimated with DeepGaze IIE, trauma-related distress, and insomnia severity. 

Misleading information robustly distorted memory: the model-estimated odds of a correct rather than incorrect response were about four times higher for non-misled than for misled details, and suggested details were more often remembered as also having been seen in the film. None of the hypothesized moderators reliably changed the size of the effect, and BIC model comparisons consistently favored the simpler models, providing no support for the predicted moderation. Detail centrality, visual salience, distress, and insomnia severity did not reliably identify which details or individuals were more susceptible. In this trauma-film dataset, the misinformation effect did not reliably depend on a detail’s centrality or image-based visual salience, suggesting that how and where a detail was encountered may matter more than how central or visually prominent it appears.