An evaluation of the use of phallometric assessment for men incarcerated for sexually offending against children in New Zealand: Past results and future directions.

Abstract

Phallometric assessment, the direct measurement of male sexual arousal in response to stimuli presented in a controlled setting, has been widely used in the assessment of sexual offenders to detect deviant sexual interests, determine treatment needs and inform risk assessments, but they have also been criticised due to a weak theoretical basis, wide variations in methodology and serious concerns around reliability and validity. From 1999 to 2007, phallometric assessments were conducted at two treatment units for incarcerated child sex offenders in New Zealand using a standardised Monarch 3.1 phallometric system which provided a database of 583 cases for analysis. This project, the only large scale analysis of phallometric data known to have been conducted using New Zealand data, was designed to explore a large number of research questions in three areas. The first area explored the factor structure of the assessments and the relationships between arousal profiles and a variety of co-existing demographic and offence related variables including age, social desirability, victim gender and victim age. The second area investigated the ability of a large number of possible phallometric indices to predict future both any sexual reconvictions and those involving children, with a particular focus on the role played by stimuli depicting teenagers. The third area investigated the prevalence and effects of the deliberate suppression of arousal, and analysed the ability of physiological measures to detect such suppression. The results of these investigations indicated that phallometric data factored according to the gender of the stimuli and could be further divided into age preferences resembling pedophila and teleiophilia. Phallometric indices consistently related to known victim gender but not victim age, suggesting that these offenders tended to target victims based on gender preferences but not age preferences to the same degree, posing questions about the relevance of diagnostic labels for age based sexual preferences. Phallometric results were demonstrated to be predictive of sexual reoffending against children and outperformed actuarial or structured dynamic variables. The best predictions were obtained using ratio and z-scored differential deviance indices from initial assessments to predict sexual reconvictions involving children in a sample of extrafamilial offenders, with a maximum AUC found of .69. Post-treatment assessments also predicted reconviction, but change scores from pre to post-treatment did not, suggesting that the practice of conducting post-treatment phallometric assessments is of little value. The investigation into the suppression of arousal found that subjects could reduce the magnitude of their arousal, but could not reduce the discriminative abilities of interpretative indices. In this sample, there was no reliable way to detect markers of suppression using GSR traces, respiration traces or the patterns in the penile traces themselves. While many of these findings support those in the existing literature, others are original contributions which extend the literature, including the use of a Principal Component Analysis on raw phallometric data, an exploration of the effects of the use of pubescent stimuli, a mathematical rather than subjective analysis of the properties of penile, GSR and respiration data in relation to suppression, the use of Receiver Operating Characteristic analysis to clarify the relationship between arousal profiles and victim preferences and an analysis of the effects of varying significance levels on the detection of male victims and the prediction of recidivism. Overall, this research extends and clarifies the phallometric literature through evidence that phallometric assessments may not provide a definitive measure of sexual interests and are not an absolute predictor of reconviction, but are the best available tool for measuring arousal patterns and could be a valuable contributor to a multimodal assessment of risk.