Dr Benjamin Spivak1, Dr Caleb Lloyd1, Dr Ariel Stone1
1Centre For Forensic Behavioural Science, Swinburne University of Technology, Melbourne, Australia
Biography:
Dr Benjamin Spivak is a Senior Lecturer at the Centre for Forensic Behavioural Science, Swinburne University of Technology. He has published over 40 papers and his current research interests are in risk assessment, cross-cultural differences and disparities in criminal justice contact and causal inference with observational data.
Abstract:
Risk assessment instruments revolutionised practitioners' and researchers' ability to predict future criminal behaviour, outperforming other prediction approaches. However, many researchers increasingly conflate capacity to predict criminal behaviour with ability to explain criminal behaviour. Whether explicitly or implicitly, it is common in the field to frame risk assessment scores as reflecting a feature or characteristic of the person being assessed (e.g., "deviancy"), or, in other words, consider risk to be an underlying psychological property within the individual. Measuring psychological constructs that explain criminal behaviour is an important and desirable goal, but risk instruments are not well-suited to this goal. Specifically, the key properties that define valid measurement of psychological features are not found in risk instruments, including the properties of order and additivity and clear theoretical separation between the attribute being measured and the measure itself. In this talk, we will challenge interpretation of risk scores as measurements of psychological constructs and discuss why this is important. We will present the criteria required to make claims of construct measurement and the research designs that could achieve this. Finally, we will discuss why this question and its answer matters, highlighting important implications for how people in the field should interpret and use of risk assessment scores.