CHAPTER FIVE

CHAPTER FIVE -

MEASURING EMPATHY IN SEXUAL OFFENDERS

KARL HANSON

Empathy involves two distinct elements:

	perspective-taking (being able to see and understand another's point of view - noticing that they are suffering)
	and sympathy (an emotional reaction to the other's suffering).

This is a two stage process in that you must notice that the other is suffering before you can react emotionally to it.

There are at least three possible reactions to noticing that another is suffering, all of which may be present in sex offenders,

	a sympathetic reaction, involving feelings of compassion care or pity;
	an unhelpful reaction involving distress, fear, avoidance or withdrawal; and
	an anti-social reaction involving anger, wanting vengeance, a feeling of superiority or power, or even sexual arousal.

Empathy may be thought of as a stable characteristic or it might be intermittent depending on the person's state. Someone might be less empathic when they were distressed themselves, or when they were sexually aroused. Or their empathy might be specific to particular situations or only triggered by particular cues. In practice, sex offender's empathy can vary in all these ways.

It is probably useful then to think of someone as having both

	a certain limited capacity for empathy (with some people having a greater capacity for empathy than others)
	but also to think of them as having a habitual level of empathy - the amount of empathy they typically display - in various situations.

A number of self-report measures of empathy have been developed. These actually measure rather different things. They cannot be thought of as interchangeable.

Mehrebian and Epstein (1972)'s questionnaire assesses the degree to which you are affected by other's emotions,

Hogan (1969)'s questionnaire seems be measuring more a kind of social competence, an ability to understand others.

Davis (1983)'s questionnaire assesses four aspects of empathy:

	perspective-taking (the habit of trying to see other people's point of view);
	emotional involvement in fantasy;
	empathic concern (sympathy with other people who are in trouble); and
	personal distress (being upset when other's are in difficulty).

All self-report measures have limitations. They assess slightly different constructs and may not be focusing on the kind of empathy which is relevant to sex offenders. They depend on the subject being aware of their empathic reactions. And they depend on the subject being prepared to candidly report their reactions. Perhaps because of these limitations most studies using general empathy questionnaires have failed to find a difference between sex offenders and non-offenders in their self-reported empathy.

Given the imitations of questionnaire measures of empathy, many people turn to interviews to assess empathy. The interviewer may attend to the subject's general manner and to the general attitudes expressed. The interviewer may compare the account the offender gives of the offence either to what the victim said or to what their common sense suggests is plausible. Where there are discrepancies this will be taken to indicate a lack of empathy.

The perspective-taking aspect of empathy is easier to assess. The basic idea is to get the subject to make judgements whose accuracy can be objectively assessed. Two methods have been commonly used to do this:

	presenting stimuli through a video-tape in which actors depict particular reactions or
	presenting written descriptions of scenarios in which the reactions of the people involved have to be predicted.

Assessing the a subject's accuracy in making judgements does of course depend on having some method for defining what will count as "right answers".

Nevertheless, despite this problem this approach has produced good results. Several studies have found that sex offenders show poorer perspective-taking on tasks which are relevant to their offending

(e.g. Lipton, McDonel and McFall, 1987; Stermac and Segal, 1989; Beckett, Beech, Fisher and Fordham, 1994).

Two new scenario-based measure have been developed specifically for the assessment of sex offenders by Hanson and Scott.

	The Empathy for Women test has 15 items each of which depicts a sexual interaction between a man and a woman. The subject is asked to judge the woman's emotional response. "Right Answers" were defined on the basis of the responses given by two samples of women (127 adults from a community sample; 100 university women).
	The Child Empathy Test has 16 items each of which depicts an interaction between an adult and a child. Some of the adults are described as behaving in what is actually a sexually intrusive way. Again subjects are asked to judge the child's response and their predictions compared to the consensus of 15 child sexual abuse experts.

In both tests the scenarios are chosen so that the correct answer may be very negative, mildly negative, or mildly positive. Thus the subjects can make too [two] kinds of error: they may attribute improbably negative reactions to the woman/child or they may attribute improbably positive reactions. The former might be produced by an offender who was trying [to] fake an empathic response whilst the latter would indicate a failure to notice how abusive some behaviour was.

These measures were administered to

	a sample of sex offenders (N=8) drawn from a number of institutions in Canada,
	to male prisoners who were not sex offenders (62),
	to male students (76),
	and to a general sample of community males (120).

All subjects were asked whether they had engaged in various sex offences.

Nineteen percent of the convicted sex offenders denied having committed any sex offence.

Interestingly though, just over a quarter of the community sample (28 %) admitted to have committed at least one sex offence

	(26 admitted rape,
	14 child molesting, and
	9 both rape and child-molesting).

The convicted sex offenders similarly admitted to rape in 21 cases, child molestation in 66 cases, and both in 39 cases (with 23 denying).

In these samples the new measures proved to only have limited internal consistency.

	The Child Empathy Test in particular only had an alpha coefficient reliability of 0.3 while
	the Empathy for Women Test had an alpha coefficient of 0.5.

These low internal consistency coefficients probably reflect the fact that the two kinds of error were negatively correlated with each other so that someone who systematically underestimated (say) a child's distress was not likely to make the kind of error involved in over-estimating distress.

The two measures proved to be largely independent of an established measure of the tendency to fake good, the Marlowe-Crowne Social Desirability scale, though there was a small trend for social desirable responding to be associated (r=0.2) with the sort of errors involved in over-estimating a woman's distress.

The measures did have some correlations with a well known measure of cognitive distortions, the Abel and Becker scale, both of the new measures correlated just under 0, with the Abel and Becker scale indicating only a weak relationship. There was also some suggestion that subjects with poor vocabularies were more likely to make errors, though interestingly only the kind of error involved in under-estimating distress.

Comparisons between sex offenders and control subjects were complicated by the fact that some of the offenders had received treatment.

On the Child Empathy Test untreated sex offenders seemed to score worse than either treated sex offenders or the comparison group though the difference was not large, amounting to only one less item correctly answered by the untreated sex offenders on the average. Interestingly though, if incest offenders were distinguished from the others and then they seemed to show a very specific pattern of errors, tending to make their mistakes just on the items which paralleled the offences they had actually committed.

A somewhat different pattern was apparent for the Empathy for Women Test. Here all groups of sex offenders, whether convicted or not, and whether child-molesters or rapists, tended to score worse than any of the non-sex offender groups. And amongst the convicted sex offenders, those who had been under the influence of drink, or who had used force in their offence, tended to make fewer errors on this test. Suggesting perhaps that the sober, less violent sex offences were committed by people who failed to recognise their victim's distress whilst the more violent or drunk offenders either did not care about their victim's distress or found some justification for ignoring it.

Overall these results are moderately encouraging suggesting that there is promise in the scenario approach. It is clear though, both that the tests could be improved and that this aspect of empathy, perspective-taking is only part of the story. Clearly different kinds of emotional reactions to a potential victim's distress must also play a pact, as well, no doubt, as other factors.

Meanwhile there are at least three ways in which scenario measures could be improved. Their biggest limitation at the moment is probably their low reliability.

	In principle this could be overcome by increasing the number of scenarios but this might lead to subject fatigue. That might be avoided if scenarios were made more interesting by presenting them though audio or video-tapes, though these might be hard to construct.
	Another possibility would be to keep the number of scenarios constant but to ask the subject to predict more aspects of the victim's response.
	Finally, there is the possibility of developing a procedure for individually targeting scenarios so that each subject is asked to make judgements about the kinds of scenario which is closest to their offence.

REFERENCES

Beckett, R., Beech, A., Fisher, D., & Fordham, A.S. (1994) ' Community-based treatment for sex offenders - an evaluation of seven treatment programmes. London: Home Office.

Davis, M.H. (1983) Measuring individual differences in empathy: evidence for a multi-dimensional Approach. 'Journal of Personality arid Social Psychology' 44, 113-126.

Hogan, R. (1969) Development of an empathy scale. 'Journal of Consulting and Clinical Psychology.' 33. 307-3I6.

Lipton, D.N., McDonel, E.C , & McFall, R.M. (1987) Heterosexual perception in rapists. 'Journal of Consulting and Clinical Psychology.' 55, 17-21.

Mehrebian, A., & Epstein, N~A (1973) A measure of emotional empathy. 'Journal of Personality,
40, 525-543.

Stermac, L E & Segal, Z. V. (1989) Adult sexual contact with children: an examination of cognitive factors. ' Behaviour Therapy.' 20, 573-584.