Developing Attributes and Attribute-Levels for a Discrete-Choice Experiment: An Example for Interventions of Impulsive Violent Offenders

Background Discrete-Choice Experiments (DCEs) are used to assess the strength of preferences and value of interventions, but researchers using the method have been criticised for failing to either undertake or publish the rigorous research for selecting the necessary attributes and levels. The aim of this study was to elicit attributes to inform a DCE to assess societal and offenders’ preferences for, and value of, treatment of impulsive-violent offenders. In doing so, this paper thoroughly describes the process and methods used in developing the DCE attributes and levels. Methods Four techniques were used to derive the final list of attributes and levels: (1) a narrative literature review to derive conceptual attributes; (2) seven focus group discussions (FGDs) comprising 25 participants including offenders and the general public and one in-depth interview with an offender’s family member to generate contextual attributes; (3) priority-setting methods of voting and ranking to indicate participants’ attributes of preference; (4) a Delphi method consensus exercise with 13 experts from the justice health space to generate the final list of attributes. Results Following the literature review and qualitative data collection, 23 attributes were refined to eight using the Delphi method. These were: treatment effectiveness, location and continuity of treatment, treatment type, treatment provider, voluntary participation, flexibility of appointments, treatment of co-morbidities and cost. Conclusion Society and offenders identified similar characteristics of treatment programs as being important. The mixed methods approach described in this manuscript contributes to the existing limited methodological literature in DCE attribute development.