Eight Principles Aim to Turn Crime Storytelling into Rigorous Scripting

Criminal acts rarely unfold as single moments. For three decades, researchers have captured these sequences with crime scripts, step-by-step procedural models borrowed from cognitive psychology.

Crime script analysis, introduced by Derek Cornish in 1994, treats crime not as an instantaneous act but as a process with events happening before and after the principal offence. By mapping the procedural steps, decision points and enabling conditions of an offence, analysts can identify pinch points where tailored interventions might disrupt the sequence. The method has moved steadily toward the mainstream: the UK College of Policing has produced guidance videos on developing crime scripts, and the United Nations recommends their use against crimes such as procurement fraud and wildlife crime. Published applications now span drug production, sexual offending, cyber-enabled fraud, organised crime markets and irregular migration.

Yet according to Borrion and his colleagues, including Paul Ekblom and Alina Ristea, the methodological side of script construction has lagged behind its empirical popularity. Most of today’s knowledge about the make-up of crime scripts still traces back to Cornish’s original papers, which recommend a template called the universal script, composed of nine generalised stages known as scenes: preparation, entry, precondition, instrumental precondition, instrumental initiation, instrumental actualisation, doing, post-condition and exit. The authors note persistent confusion about this template, with some later papers reporting seven stages and others eight, and one widely cited guidance document adding a tenth scene called Aftermath.

The new paper grounds its eight principles in two concepts drawn from systems theory and business process modelling: states and activities. A state represents the characteristics of a system at a point in time, such as an offender’s location being outside a dwelling. An activity describes the decisions and interactions over a period that explain what happens between two states, such as the offender entering the dwelling. Framed this way, every procedural unit in a script should represent a transition from one situation to another, usually caused by an activity. Even internal processes like criminal intent or cognitive assessment count as activities, the authors argue, because they can change states.

The eight principles, all beginning with the letter C, address Constitution, Chronology, Composition, Cardinality, Categorisation, Completeness, Clarity and Consistency. Constitution requires each stage and step to represent a transition between two situations. Chronology demands that stages and steps follow the order in which they are believed to have occurred, a central feature given that situational crime prevention aims to intervene before the principal act takes place. Composition requires each stage to be more abstract or longer-lasting than the steps it contains, and each step more specific or shorter than its parent stage. Cardinality allows a stage to contain many steps but insists each step belongs to one and only one stage, preventing duplicated effort or overlooked content when multiple analysts share a script.

The remaining principles govern grouping and completeness. Categorisation requires steps to be placed under the stage that most accurately reflects their timing and influence on the unfolding situation, a common failure being an action placed in the wrong row of a tabular script. Completeness demands no missing stages or steps, because an omitted step means the corresponding pinch points may never be identified, and, if a script represents multiple events, an omission risks being misread as evidence that an action did not occur. Clarity requires that readers understand the situations before and after each step, specifying at least the start and end states, since a movement verb like walk conveys motion but not trajectory. Consistency, the final principle, holds that conceptually similar stages should be decomposed using the same rules, so that improvements to a shared rule benefit every comparable step at once.

The authors illustrate the principles with abstract diagrams of a hypothetical three-stage script and nine track variants, only one of which adheres to all eight. Each variant violates a different principle: one reverses the chronological order of two steps, another lets a step stretch across two stages, another omits a step entirely, and another decomposes two nearly identical stages, such as assaulting a first and a second target, into very different sequences of steps. The principles are theoretically justified through such constructive counterexamples rather than through new empirical data, and the authors are explicit that they remain in need of systematic empirical assessment.

A substantial part of the paper examines what the authors see as unexamined design choices embedded in existing templates. They argue that several features often treated as foundational are pragmatic decisions shaped by project-specific constraints. Leddo and Abelson, whose work underpins the universal script, needed ten scripts of equal length for a psychology experiment and split an original seven-stage model into more scenes to fit their purposes. Tompson and Chainey later condensed the nine scenes into four temporal categories because information on illegal waste activity, their case study, was scarce. Cornish himself acknowledged difficulties distinguishing between preconditions and instrumental preconditions. The paper also questions the standard four levels of script abstraction, from metascript down to track, arguing that this hierarchy is arbitrary and that intermediate categories can always be inserted.

These choices matter in practice, the authors contend. Reducing scenes to purely temporal classifications, as later template revisions did, strips away the functional cues that help analysts identify and organise relevant information, probably increasing the risk of information omission. Tying scenes to physical settings undermines recursive decomposition, because entry and exit scenes lose their meaning once an actor is already inside the setting. The authors point to published acknowledgements, including one in this journal on wildlife trafficking, that no fixed correspondence exists between the stages of real crimes and the functions of the universal script.

The paper concludes with a call to move beyond templates altogether, toward a more flexible scripting method that retains the chronological and functional cues of the universal script but can be applied at any level of abstraction and resolution. The eight principles are offered as a meta-framework for evaluating existing methods and guiding the design of new ones, both by assessing method designs against each principle and by examining the scripts those methods produce, though the authors caution that output quality alone cannot confirm a method’s validity. They also acknowledge limitations: the principles were developed for simple linear scripts involving a single offender and may need relaxation for scripts with multiple interacting actors, branching decisions or feedback loops, and it remains unclear at which point in the scripting process they are best introduced. Empirical work to test and refine them, including their use as prompts for large language model applications, is already underway.

Subject of Research: Eight principles for improving the decomposition and aggregation of crime scripts in crime script analysis.

Article Title: From storytelling to scripting: eight principles for deconstructing crime events

Article References: Borrion, H., Ekblom, P., Bish, A., Fan, C., Gómez-Quintero, J., Chen, P., Toubaline, S., Ristea, A., & Niri, M. (2026). From storytelling to scripting: eight principles for deconstructing crime events. Crime Science, 15(1), Article 30. https://doi.org/10.1186/s40163-026-00286-w

Image Credits: AI Generated

DOI: 10.1186/s40163-026-00286-w

Keywords: crime script analysis, crime science, crime prevention, situational crime prevention, universal script, procedural decomposition, offender modelling, methodology, states and activities, Derek Cornish, event schema, script quality