KOINOS MAGAZINE #20 (1998/4)
Radical Reconsideration of the Concept of Child Sexual Abuse
New Findings by Bauserman, Rind and Tromovitch
Earlier in the present volume of Koinos (issues 17 and 18) we exhaustively reported on research by the American scholars, Dr. Bruce Rind and Dr. Robert Bauserman. Recently, since joining forces with Dr. Philip Tromovitch, they have published two new scientific articles in which they provide statistical meta-analyses (research on research) regarding sexual experiences of minors with older persons. In them they test the scientific accuracy of the concept of Child Sexual Abuse (CSA) and four conceptions or implications customarily associated with it. In particular, they examined the connection made between CSA experiences and psychological problems later in life. Their conclusions are startling.
The two new publications parallel one another with regard to the research questions, methodology and conclusions, and differ in the research samples from which they draw the data they analyze. Both papers involve non-clinical and non-juridical populations, after Bauserman and Rind in an earlier literature review (discussed in Koinos 17) had established that these differ in essential ways from clinical populations whose subjects were interviewed while in therapy for CSA. The second investigation, from 1997, concentrates on seven studies using samples drawn from the general population (in addition to four from America, one each from Great Britain, Canada and Spain). The third research, published in 1998, examines 59 surveys of populations of American college students. In relation to this, Europeans must realize that, unlike the practice here, half the adult population in the United States has gone on to participate in education beyond the secondary school level. Therefore there were not important differences to be found with surveys of general population samples; the age at which the students were surveyed also had no measurable influence. The advantage of the use of the college research is that it provided extensive data, including on psychological well-being, and offered information from many male respondents.
Rind, Tromovitch and Bauserman begin both articles by showing that the concept of CSA is defined in various ways, but generally quite broadly. It involves not only children, but also adolescents, often to the age of sixteen or even eighteen, and it also includes sexual experiences which did not involve force, if these involved incest or a partner who was considerably older - generally by five or more years. Despite expressing initial reservations - in scientific terms, ‘abuse’ assumes the intention of inflicting damage - for the sake of convenience they continue to use the term CSA.
They next follow this by looking at the four characteristics which are consistently assumed for CSA in both professional and popular literature:
All of these characteristics have been developed from research on clinical populations; Bauserman and Rind had already previously shown (see Koinos 17) that the fourth of these characteristics is not valid for non-clinical populations.
In order to make the vague term ‘harm’ specific enough to use in comparisons, meta-analysts work with a list of types of diminished well-being, often cited in self-reports and employed in standard surveys, arranged on the basis of the statistics used into eighteen more or less standard symptoms: alcohol abuse, anxiety, depression, dissociation (multiple personality syndrome), eating disorders, hostility, interpersonal sensitivity, loss of control, obsessive/compulsive behavior, paranoia, phobia, psychotic symptoms (aural or visual hallucinations), diminished self-esteem, problems with sexual adjustment, social maladjustment, psychosomatic symptoms, suicidal impulses and, lastly, the general category of ‘wide adjustment problems’.
Before discussing the relationship between CSA experiences and these forms of psychological harm, the authors warn that despite the statistical terms ‘effect’ and ‘effect size’ being constantly used, a connection implying a cause-effect relationship is still not demonstrable. Many researchers assume that cause-effect relationship, either silently, as a matter of fact, or explicitly state that is the case.
I find it remarkable that Rind and Bauserman do not qualify either their own earlier study, or that of Constantine from 1981, which was their leading model, as meta-analyses. They instead characterize these investigations into previous research as qualitative literature surveys, of which there are others (albeit mostly exclusively of research on clinical populations), and reserve the term meta-analysis for quantitative, statistical treatments of previous statistical research results. With these they reach a much higher level of precision and demonstrable reliability, but unfortunately this seriously reduces readability. Anyone who is not at home with the language of statistics, and when encountering ‘Fisher Z’, for instance, can only associate that with a reasonably successful pop group from around 1980, has no choice but to skip over large parts of the argumentation as incomprehensible.
In this year's article, two recent, previous meta-analyses are dealt with more specifically: those by Jumper, from 1995, and by Neuman, et al, from 1996. Neuman's meta-analysis examined 38 studies, exclusively involving women, half of which came from clinical settings. Here a small to moderate relation was found between CSA and diminished well-being, in which the non-clinical populations clearly scored more favorably. The age of the subjects at the time of the investigation produced no differences. Jumper, in her analysis of 26 studies, of which half were clinical and in which 83% of the subjects were women, came to an entirely different conclusion. In her analysis the general population samples paralleled the samples from clinical and juridical settings, and college respondents were an exception with a much lower relation between CSA and diminished well-being. It was merely that women students with CSA experiences reported reduced self-confidence.
In what is certainly an entertaining passage, the authors report that, upon review, Jumper's calculations don't appear to tally. An expert meta-analyst they consulted came to the same conclusion. Alas, Jumper responded that her data and calculations were in storage in a different part of the country, and suggested that Bauserman, Rind and Tromovitch proceed with publication, simply reporting that they were unable to resolve the discrepancies with her. That is exactly what they do, in a deadly footnote. The recalculated data from Jumper's analysis indicate that, entirely in line with Neuman, et al, and Rind, et al, her general population samples parallel college samples, and not clinical samples, and, like Rind, et al, the relation between CSA and diminished well-being is even weaker for non-clinical male respondents than for non-clinical female respondents.
To be included in the 1997 meta-analysis, general population studies had to fulfill four conditions: 1) present quantitative data; 2) report on effects on psychological or sexological adjustment; 3) report data for men and women separately; 4) have included a control group. This yielded seven studies from the years 1984 through 1995, with the number of respondents in each ranging from slightly over 300 to over 3000. In Laumann's American research (1994), 70% of the women reported negative effects of CSA, as opposed to 46% of the men. The British research by Baker and Duncan (1985) yielded comparable figures: 37% of the men reported harm of any kind (only 4% identified the harm as lasting), as compared to 64% of the women (13% lasting). On the other hand, in research in Canada by Badgley (1984), only 4% of the men reported physical harm and 7% psychological harm. Among women these percentages were 20% and 24%. But Badgley asked only about first experiences, and short-term harm. However, that reports of harm among male respondents are less negative is a general tendency. This is also reflected in the questions about negative feelings in Lopez's 1995 research in Spain, and American studies by Bigler (1992), and Boney-McCoy/Finkelhor (1995). In the Lopez study men also reported more positive feelings (pleasure, satisfaction) in relation to their CSA experiences. In regard to this, it is also worth noting that women on the average reported different sorts of CSA experiences than man: incest was three to four times more prevalent, violence was much more common, and their experiences were clearly more frequently from before the onset of puberty.
It appears that harm, gaged according to the eighteen symptoms of diminished well-being listed above, was much less measurable from the quantitative data than the self-reports in the immediately preceding paragraph might suggest. Only with regard to the point of ‘problems with sexual adjustment’ did Laumann find any significant differences with his control group. But over against that stood the fact that the members of the CSA group appeared to be more sexually active, and to participate in more varied forms of sexual activity, than those of the control group; this could explain why they also encountered more problems in this area. Lopez and (in a 1989 study) Finkelhor also found minimal differences with regard to ‘problems with sexual adjustment’, respectively in attitudes about CSA questions (presupposing that a sexual trauma leads to more negative attitudes regarding this subject). Only in the Boney-McCoy research is the effect size among women - and to a lesser extent among men - much greater. This can be explained because in this study inquiries were made exclusively concerning unwanted CSA experiences. In meta-analysis of the relation between CSA and symptoms of diminished well-being, there appeared on the average to be an effect size of only 1% (Boney-McCoy not included), and still less than that for males. That is to be compared with 10%, however, among clinical samples. In closing, the meta-analysts discuss another study, one by Ageton (U.S., 1988), which had not met their criteria as there were no men in the sample. From this study, it appears that there are no differences in generalized measures of well-being between women who as children suffered from non-sexual abuse, neglect or negative environmental factors such as criminality or drug abuse in the home, and women who suffered all these plus CSA experiences.
Nature of the experiences
Keeping in mind this last, the fact that there are other possible causes for diminished well-being besides CSA, Rind and his colleagues went to work on a meta-analysis of 59 studies (all but two from the years 1983-1995) from college populations, with the impressive total of 35,700 respondents (of whom 22,000 were women), with between 69 and 3200 respondents in each study. The criteria which these studies had to fulfill were: 1) the sample be of college students, and include a control group who had not had CSA experiences; 2) that there be a distinct CSA group rather than a general ‘abused’ group including subjects without a history of CSA; 3) that the study report on at least one of the eighteen symptoms listed above; 4) that it provide sufficient data to compute effect size.
The definitions of CSA used in the studies were rather diverse, but 70% of them also included experiences willingly engaged in by the subject if the partner had been considerably (at least five or eight years) older; 73% also included non-contact CSA (exhibitionism on the part of either party), and 75% also included more mature adolescents (the upper age limit being around fifteen). Working from these wide definitions, 27% of all women and 14% of all men reported experiences with CSA. The nature of the experiences is shown in the following table.
Several circumstances which might have an effect on the severity of the experiences are shown in the following table (NB: ‘Incest’ is limited to CSA with a parent, step-parent, grandparent or considerably older brother or sister):
Just as in the case of the general population samples, there was a small effect size, an average of 1% over seventeen of the eighteen symptoms. (In the case of ‘loss of control’ there was no difference at all.) Two of the 59 studies appeared to show a different pattern; in both cases they involved populations composed chiefly of women with incest experiences. The men with CSA experiences involving force or threat of force scored as significantly worse off in comparison to both men with CSA experiences without force or threat, and the control group. They scored about the same as women with CSA experiences, whether these had involved force or not. No differences were found when results were examined in terms of the differences in ages at the time of the research, or the academic field being studied by the respondents. Strangely enough, in the published research reports the effect size was somewhat greater than in the unpublished ones. In the case of incest and, particularly, violence, the authors found a greater effect size, all along the line. This was not the case when the variables examined were penetration or its absence, higher or lower frequency of the occurrence of CSA, or the total duration of the CSA experience.
In the case of ten female and eleven male populations (1421 and 606 persons), questions were asked regarding immediate reactions at the time of the CSA. In seven female and three male populations (510 and 118 persons), questions were also asked about current reflections on the experience. The following table gives a survey of the results:
With regard to self-reported effects (eleven populations), an average of 13% of the women and 8,5% of the men with CSA experiences reported lasting negative effects, generally having to do with stress, emotional harm or problems with their sexual life. With regard to other symptoms, the percentages were much lower. In particular, a clearly distinguishable portion of the women respondents reported temporary negative reactions. All across the board, in respect to well-being, men scored more favorably with regard to CSA experiences than did women.
As might already have been somewhat expected from Ageton's research involving a general female population, the influence of non-sexual child abuse, neglect and unfavorable family environments appears clearly in meta-analysis in which CSA experiences, non-sexual experiences of this sort and symptoms of diminished well-being are statistically correlated with one another. First, it appears that a higher percentage of the students with CSA experiences had also experienced problematic family environments (already problematic before the onset of CSA) than was the case for control groups without CSA experiences. The effect sizes of these unfavorable family environments were moderately large for all eighteen symptoms, to a degree that Rind, et al, considered the chance to be nine times greater that family environment explained the symptoms, than that the CSA was responsible. By statistical control on the fourteen studies where that was possible, in 83% of the cases the already small relation between CSA and the symptoms was no longer statistically significant once correction was made for family environment. This upheld the statistical validity of the studies.
The authors wrap up their latest meta-analysis with the conclusion that the customary term CSA, and the four characteristics assumed to accompany it, demonstrably do not apply in the case of non-clinical populations. They therefore argue that a separate, value-neutral term, ‘adult-child sex’, be adopted for willing encounters with positive reactions, and similarly ‘adult-adolescent sex’ be used to designate willing encounters with positive reactions between adults and adolescents; ‘child sexual abuse’ would then be retained for encounters in which a child did not freely participate, or in which there were negative reactions, and ‘adolescent sexual abuse’ be used for similar encounters involving adolescents. Lastly, they warn against the intermingling of moral and legal concepts with scientific concepts. In this light, they offer the closing observation that they are making no judgement about whether sexuality between adults and minors is right or wrong from a moral or legal perspective, except where moral or legal judgments are based on the mistaken supposition of psychological harm.
Rind, B. and Tromovitch, P., A Meta-Analytic Review of Findings from National Samples on Psychological Correlates of Child Sexual Abuse, in: Journal of Sex Research, Vol. 34/3, 1997, pp. 237-255.
Rind, B., Tromovitch, P. and Bauserman, R., A Meta-Analytic Examination of Assumed Properties of Child Sexual Abuse Using College Samples, in: Psychological Bulletin, Vol. 124/1, 1998, pp. 22-53.