11.4. Analysis.

Comparison 11: Psychosocial risk management ('Resettle programme') versus treatment‐as‐usual, Outcome 4: Recidivism: binary outcome (no offences vs 1 or more offences) for official offences at 2 years post release, corrected for time in the community

Recidivism: binary outcome (no offences vs 1 or more offences) for official offences at 2 years post release, corrected for time in the community
Study	Estimate	SE	95% CI	p	Comments
Nathan 2019	2.077	1.12	‐0.188 to 4.201	0.073	Statistical prediction of the zero‐inflation binary factor (no offence vs. 1 or more offences) reported by ITT Resettle group versus the control, reported by the study investigators (n = 72, 90% AsPD). The study authors stated (p 5, col 1) "When conducting zero‐inflated regression models, Mplus creates a continuous outcome measure as well as a latent binary outcome measure, since zero‐inflated measures benefit from examining whether participants who score zero versus any other value other than zero might differ in relation to the independent variables. Positive values on the binary outcome are interpreted to mean that lower values are related to greater chance of the dependent variable assuming zero values. Negative values on the binary outcome are interpreted to mean that higher values (using categorical contrast coding) on the independent variable are related to more non‐zero values on the dependent variable."