~ [Home] ~
COMPARING PHYSIOLOGICAL AND SELF-REPORT MEASURES
Much research on physical arousal to deviant sexual stimuli fails to address the question of its power versus that of other approaches. The degree to which the phallometric and other approaches tally is taken as evidence of the validity of phallometry.
Some studies compare groups of men with different offending characteristics (e.g. offenders against boys with rapists of adult women) to assess whether the known offender groups differed in their physical response to different types of deviant material. We have seen many examples of this already. Others compare the amount of erection to the deviant stimuli as measured by the machine with the offenders' self-reported arousal to the material.
Although these provide a degree of confidence about the worth of phallometry, they do not in themselves demonstrate the superiority of physiological methods over, say, questionnaires for identifying the man's deviant sexual characteristics. In practice, the questionnaire might be more effective at discriminating between the types of offender. Research needs to address the relative power of the different methods of assessment -- a sort of shoot-out between the two.
Day et at. (1989) took rapists and sexual offenders against children. One objective was to examine whether the physiological methods could distinguish between violent and non-violent offenders. To this end, police and probation reports prepared at the time of sentencing were used to assess whether the offender was violent -- a third were by the criteria of injury, verbal abuse or the use of a weapon. Penile circumference was measured using a mercury-in-rubber strain gauge in response to slides of adults and children of both sexes, audio tapes and video tapes.
Indexes were developed based on arousal to the following types of content:
Self-reported sexual behaviour was assessed using the Multiphasic Sex Inventory (Nichols and Molinder, 1984), which consists of 300
true/false questions dealing with many aspects of sexual activity.
Subscales from this deal with
Using complex statistics, it was found that the different aspects of sexual arousal as measured by phallometry correlated fairly well with self-reports of similar behaviours.
However, the amount of variation shared between the two types of measures was only 25% of the maximum possible, implying that phallometry and the questionnaire were largely measuring different things. It appears that a dimension from homosexual/non-coercive sexual preference to heterosexual/coercive sexual preference mainly accounts for the association. That is, coercion and heterosexuality were correlated.
A number of analyses were carried out in order to maximize the discrimination between offenders on the basis of type of victim.
For the child molesters, the physiological measures could distinguish offenders against boys from those against girls. The men who were aroused by material including male sex partners were more likely to have offended against boys, and arousal to rape themes identified those who had offended against girls.
Self-reports subjected to a similar analysis also discriminated. The child molestation measures were associated with known child molesting, as were the rape measures with known rape offenders. The sex of the victim of child molestation was fairly well predicted by sexual orientation (homosexuality versus heterosexuality) and preference for male children. However, having girl victims was associated with rated preference for female children and rape.
While the general impression is of the superiority of the self-report measures, especially in terms of consistency, the best evidence comes from the relative abilities of self-report and physiological measures to predict offending history.
Day et al. found, using discriminant function analysis, that self-reports correctly classified 95% of the sample by known type of offence -- rape or child molestation. Eighty-six per cent of the child molesters were correctly classified on their victim's sex, and 85% were correctly classified on the use of violence in offending. In every comparison, "hit" rates for self-report data were better than those for the corresponding physiological measure. Thus, only 74% were correctly identified as rapists or child molesters; 82% of molesters were correctly identified according to their victim's sex and 74% on the use of violence.
According to Day et al. (1989), self-reports may measure history of offending and the physiological reports tap "aspects of the potential to offend" (p. 122).
The authors fail to address the question, "why bother with either the physiological data or the self- report data?" The key data of offending is seen as the criterion by which these other aspects are judged. Why not just leave things at that?
There are a number of possibilities, of course.
One is that, clinically, the additional information may reveal problem areas that the offending history in itself is unable to
show -- offending is more extensive than arrest.
Unfortunately, appealing as this is to common sense, it remains an open question as to what aspects of a penile plethysmograph can be interpreted diagnostically in the absence of secondary confirmation by the offender through interview or other forms of self-report.
Faced with the complexities of abusers' cognitions about their offending (and the possibility of lying and denial), it is not surprising that some clinicians have sought to use apparently objective measures such as penile plethysmography. That it is probably as problematic and uncertain as any other approach to the assessment of paedophiles means that we should be wary of relying on it as a major criterion of the success of any treatment or therapy (e.g. Jenkins-Hall et at., 1989) in the absence of evidence from recidivism data