This study aims to evaluate the circumstances and evidentiary grounds underlying false or baseless rape reports. Structured as a continuation of prior machine learning research using the same corpus of rape report narratives, this study uses qualitative content analysis to examine the application of unfounded classifications in rape reports and to generate a more generalizable estimate of false and baseless reporting rates.
The study uses a unique, large data set (n = 5,638) of rape report narratives linked to sexual assault kits from one Midwestern US jurisdiction over 24 years. The authors conduct a content analysis on a random subsample (n = 122) of the data set’s unfounded cases (n = 386).
A very small proportion of cases met the criteria for false reporting (2.7%) (1.1% involving victim recantations and 1.6% determined false based on evidence). An additional 1.7% were classified as baseless. Lastly, 2.4% lacked sufficient detail to justify the unfounded designation, highlighting concerns regarding premature unfounding and inadequate documentation. Most unfounded narratives consisted of brief statements that questioned credibility or alleged falsehoods, often supported by limited evidence. These findings indicate that conflating false and baseless reports can inflate perceptions that victims fabricate allegations and obscure investigative deficiencies.
Drawing on a substantially larger, more generalizable sample, this study offers a nuanced understanding of unfounded practices spanning nearly 25 years. It provides more accurate estimates of false reporting and identifies critical areas for enhancing documentation, investigative rigor and compliance with federal guidelines.
