SATISFY EASY LADDERS: accessible machine learning to facilitate analysis of larger sensitive text data samples in criminology

Large-scale data analysis has advanced criminology, but context-rich text is largely analysed manually in small samples. Machine learning (ML) with natural language processing (NLP) allows scalable analysis, but criminological use remains limited. Adoption is hindered by sensitive administrative text with constraints on data access, governance, expertise, infrastructure, time, and small datasets. This brief report introduces a framework to support non-specialist criminologists in applying ML techniques to sensitive text data in such contexts. The framework organises established ML NLP methods into an end-to-end workflow, providing mnemonic aids, ‘SATISFY’ and ‘EASY LADDERS’, to encourage adoption. Its application is demonstrated by classifying 193 child safeguarding practice reviews (CSPRs) to identify exploitation, missing incidents, school exclusion and special educational needs and disabilities (SEND), producing new prevalence evidence, and showing that modest training datasets suffice with an explainable, human-in-the-loop approach. The framework is generalisable beyond CSPRs to other document types, enabling scalable sensitive text data analysis across criminological domains.