By Dr. Frans Gieles
In: Ipce Newsletter E 7, December 1999
In the last Newsletter, I told about the controversy concerning the Meta-Analysis of Rind et al. in chronological order. In this article, we will take a look at the counter arguments against the Meta-Analysis.
For an honest scientific debate, one has to put counter arguments in the form of a question, not in the form of a proposition. Most of the opponents have spoken in propositions, mostly with strong emotional loading and much rhetoric. In this article, I translate these propositions into questions and supply answers.
The sources of the answers were several Ipce members and several others. Thanks to everyone who sent messages to the various lists and sites. From this mass of printed messages, I could write a book, but I prefer to write an article and to get to the point.
Paul Fink, M.D., a former president of the American Psychiatric Association, said in reference to the study - "I am at a loss to understand even the premise of this work ... Scientific inquiry proceeds from real-world observations, which are then used to formulate hypotheses. The apparent hypothesis motivating this research - that child sexual abuse is not nearly as harmful as we think - cannot be justified by any interpretation of the literature."
Studies that test a hypothesis are one kind of research, there are other forms. A study can also have a question, like "how much harm does child sexual abuse do, and what do the collected data so far suggest"? Or, if you insist on a hypothesis, what about "a meta-analysis of all well-designed studies conducted will yield insight into the quantity of harm done"?
There are also exploratory studies done and ones with practical consequences. The opportunity to do a meta-analysis arises only after there has been a great deal of research on a question.
Meta-analyses are just a statistical way of determining whether there is converging evidence for a particular outcome. No single study ever answers a question conclusively. Science advances through replication and converging evidence from multiple studies of the same phenomena. Unless studies are done exactly the same as previous studies, there must be some way of resolving differences and making comparisons across them. That is what the statistical procedures involved in meta-analysis attempt to do. A meta-analysis is a less biased way of combining results across studies and figuring out what they say collectively, than previous methods are, because it applies a rule that is independent of the author's assumptions and hopes for the outcome. A meta-analysis does rely on real-world observations because it is based entirely on studies that do so.
Because the statistical techniques exist in the field independent of the researcher and what is being studied, they are less likely to be the source of bias. People get distracted by such ideas as "there are lies, damned lies, and statistics" into believing that anyone can lie with statistics. That is only so when the audience is ignorant of how statistics work. There are correct and incorrect uses of statistics and these can be independently verified by readers (and by the original peer reviewers).
Opponents have said that the meta-analysis does not report about facts because of the use of self-reports that only tell what the students have said instead of what really has happened.
The use of self-reports is a wide spread and accepted practice in all social sciences. Self-reports have been used in all the studies that the meta-analysis reviewed. None of the opponents criticised the reviewed studies as long as the self-reports spoke about many harms and problems as a result of the CSA (child sexual abuse). They have ignored reports that were positive about the CSA experience. In the meta-analyses, however, these positive reports are mentioned. Just because the opponents do not like positive reports, they oppose the use of self-reports. The opponents, who always believed the negative self-reports,, ignore the voice of the students who reported neutral or positive effects. In the self-reported reaction data, 156 women (11% of 1421) reported positive reactions, 256 (18%) neutral reactions; 224 men (37% of 606) reported positive reactions, 176 (29%) neutral reactions (Table 7). The opponents do not believe these 812 people; they only believe the 1223 people who reported negative reactions.
The authors pretend to speak about the population in general, but they have only analysed self-reports by students, who are only a part of the general population.
The opponents have always believed that the conclusions of studies with a clinical or legal sample were representative for the population in general, as long as the conclusions were in terms of many harms caused by CSA. It is clear that clinical and legal samples are not representative of the population in general.
The main findings in the college sample meta-analysis were identical to the main findings of a previous study (Rind & Tromovitch, 1997) based on national probability samples. Thus on empirical grounds, it appears that the college sample findings are representative.
Students, so said the opponents, are too young to experience in full the negative effects of CSA. Their youth gives them some resistance; they will experience the negative effects later on when they are married.
Opponents who use this argument clearly did not read the meta-analysis. The authors say on page 27 of the meta-analysis: "A possible shortcoming of focusing on the college population is that college students may be too young for symptoms to have appeared, or they may be better able to cope with CSA stresses than persons in other populations (Jumper, 1995). However, younger and older adults did not differ in CSA-adjustment relations in Neumann et al.'s (1996) meta-analysis. Furthermore, mean effect sizes from college samples, as reported by Jumper, were similar to those from national samples (Rind & Tromovitch, 1997), nonclinical samples (Neumann et al., 1996), and community samples (Jumper, 1995 , after corrections). Therefore, the argument that college students are better able to cope and thus present fewer adverse reactions than people in other nonclinical populations lacks empirical support." See also page 42 of the meta-analysis. For the note and the references, see the meta-analysis.
For example, the FRC wrote: ". . .the authors loaded their analysis with data involving primarily mild adult-child interactions involving little or no physical contact. . .from which one would expect to find less permanent harm."
No "loading" has taken place because Rind et al. were very clear that they used every college study of CSA that is known to exist in the English language. When one takes everything, one, by definition, cannot have loaded more of one thing than another. One might also make the point that societal definitions of "sexual abuse" are now so broad that they fail to differentiate among gradations anyway; certainly social conservatives, such as those at FRC, have rarely pursued such subtleties before.
Rind et al. had no choice but to include all the studies in their meta-analysis, even with all their divergent definitions of CSA. If they hadn't, they would have been accused of selectivity. Instead, they're accused of over-inclusion. One has to wonder, though, where all the right-wing critics were when the earlier studies, with their sometimes over-broad definitions of CSA, came out. Why did these critics not complain back then? It is more than a little hypocritical for the right-wing to claim - as they always used to - that all forms of CSA are very harmful, and only now turn around and accuse others of wrongfully making the same assumption.
As far as I know, there are no studies that only use severe definitions of CSA. Thus, a meta-analysis can not be done for lack of studies to use as data.
Nevertheless, let’s have a look at page 30 and table 1 of the meta-analysis. Let’s look at the differences between those studies that had only contact sex, and those that had some non-contact experiences as well. Note that these categories are overlapping; for example, some of the people who experienced exhibitionism may have also experienced intercourse, and vice-versa. Sixteen of the available 76 samples were exclusively contact sex; the remaining 58 were varying combinations of contact and non-contact. Someone has calculated the median effect sizes for both groups: .135 for contact, .10 for combination; certainly not a dramatic difference.
The FRC says in a press release, May 20 1999, as Laura Schlesinger had said before based on a letter from Dr. Fink: "Of the 59 studies included in the analysis, 60% of the data are drawn from one single study done over 40 years ago,"
The authors noted that their main findings, which evaluated lingering symptoms, did not use the disputed data at all in any of the correlation effect size analyses because Landis didn’t give any correlational data. The authors added that omitting the Landis data altogether would not significantly change the other results.
By the way, the 1.496 cases in the Landis study are 4% of the total of 35,703, not 60%. Only in table 8, the self-reported effects p 37), Landis’ cases are 746 of the 1191, that’s 62%. In table 7, Landis’ 676 cases are 33% of the total of 2017 cases. But who says that Landis’ cases are 60% of the total meta-analysis, takes a pars pro toto or simply tells a lie.
The main average effect size calculation was based on a sample size of 15,912 participants. The 676 students in Landis' sample would have amounted to only 4.2% of that if Landis had done a correlation in his study, which he didn't. Furthermore, the results of this analysis (r = .10 for females and r = .07 for males, very weak correlations) are identical to results obtained from a separate meta-analysis using national probability samples, all of which are relatively recent.
Out of 59 studies, only 3 were done before 1984 (Landis, 1956; Finkelhor, 1979; and one from 1981). Only 7 were done before 1987, that's 52 studies from 1987 or later.
Out of eight different sets of statistical analyses performed, each involving numerous calculations, the primary calculation in which Landis' numbers were used was in the self-reported reactions table (table 7). Ironically, if Landis' data are removed and the figures recalculated, boys' reactions average 50% positive, up from 37%. Girls' reactions without Landis' data give an average of 15.5% positive, up from 11%. These calculations include weighting for sample size, in keeping with proper method.
The FRC says: "twenty-three of the ‘usable studies’ were ‘unpublished’."
The vast majority of the unpublished studies are doctoral theses. Including these in meta-analyses is standard practice. The other two unpublished studies were Master’s theses. It should be pointed out that Rind et al. have found that there is a slight bias in favor of publishing studies that find harm; they are more likely to be published than studies which find less or no harm.
By the way, the original claims that "CSA" was harmful to boys irrespective of how they felt about it, promulgated especially by David Finkelhor, John Briere, and their respective collaborators, relied heavily upon unpublished dissertations and a range of other unpublished studies.
The authors say in their November-6th talk (p. 9): "The unpublished data we included came almost entirely from doctoral dissertations, which, as most academics know, are generally well supervised by a group of Ph.D.s from design, through data collection, to presentation. Even more importantly, in our review we actually compared the CSA-symptom association in the unpublished and published studies a fact that has been completely ignored by our critics. The mean associations were both small (unpublished data r = .08; published data r = .11) and were not statistically significantly different, and certainly not different in a practical sense. In short, the unpublished and the published data were " telling the same story.
Most of the opponents said the meta-analysis was flawed or was garbage and full of biases.
Ray Fowler, Ph. D., writes at May 25, representing the APA: "Because the article has attracted so much attention, we have carefully reviewed the process by which it was approved for publication and the soundness of the methodology and analysis. This study passed the journal's rigorous peer review process and has, since the controversy, been reviewed again by an expert in statistical analysis who affirmed that it meets current standards and that the methodology, which is widely used by the National Institutes of Health (NIH) to develop guidelines, is sound."
Unless these words, APA asked the AAAS (American Association for the Advancement of science) perform another independent review of the Meta-Analysis. This Committee stated that " We see no reason to second-guess the process of peer review used by the APA journal in its decision to publish" and went on to say that "we saw no clear evidence of improper application of methodology or other questionable practices on the part of the article's authors."
The FRC wrote on May 20: "In an ethical and legal perspective, consent means informed consent which implies a certain level of knowledge and life experience. […] That is why the authors' premise and the basis for the conclusions in their study, that the children in the studies "consented" or were "willing" participants in adult-child sex must be rejected, and why this study should be rejected as "junk science." Whatever the children's perceptions were, we know, […] that the children were not capable of legally, ethically, morally, mentally, or emotionally consenting to sex with anybody, much less an adult. Thus, what we are left with is a study that should never have seen the light of day, much less publication in a professional journal."
The idea that children can consent to sex with adults is not a premise of this study. "Our study was based on the premise that there was a need to review research on the effects of, and reactions to, CSA experiences in college populations." (The authors in their statement of May 12.) If the reviewed studies have made distinctions between wanted and unwanted sexual experiences, a meta-analysis can make the same distinctions. Several studies did so, thus the meta-analysis did. Several authors before Rind et al. have accepted that their participants told them that their experiences were wanted in a certain percentage of the cases. A scientist has to accept this as a matter of fact. The fact is: ‘the participants told me…’. As the authors said it: "In the research we reviewed on the effects of CSA, ‘consent’ has meant the victim's own perception of his or her level of participation - from being forced to willingness - because this is known to affect a victim's reaction to the experience. Given that our study is a review of dozens of other studies, many of which explicitly examined how victim's own perceptions of their level of participation affect outcomes, it was appropriate for us to examine this factor as well." Thus, ‘consent’ was not a premise, nor a central concept, but one of the factors that could vary the participant’s perception of the CSA event.
If one rejects the possibility of willingness, one should reject every study that finds a difference between willing and unwilling experiences. But if consent to sex - informed or not - is impossible for children to give to adults, one has to then wonder why it makes such a dramatic difference in outcomes. Though a self-perceived level of consent may be of no interest to FRC, the meta-analysis demonstrates that the self-perceived level of consent makes a huge difference to actual children.
The FRC statement cited above, "Whatever the children's perceptions were, we know that the children were not capable…" is quite clear. What children say and feel is not important because FRC knows better.
If the children say they were harmed, they must be believed. NAMBLA has to accept the facts. But if they say they were not, they must still be believed. FRC has to also accept the facts. Otherwise, the children are just being used.
In the words of the authors in their November-6th paper (p 13.): " In Webster's 3rd New International Dictionary, the first definition of consent is: "compliance or approval especially of what is done or proposed by another." This definition can be termed "simple consent," of which children and adolescents are certainly capable; in fact, ethical guidelines for research with adolescents and children typically require researchers to obtain the agreement or assent of the participant. The second definition is: "capable, deliberate, and voluntary agreement to or concurrence in some act or purpose implying physical and mental power and free action." This second definition is "informed consent," which the law takes into account and which is also the typical ethical and social definition. Thus, the term "consent" clearly does not always or inevitably imply informed consent. More important from a scientific view is the value of simple consent in discriminating reactions or outcomes. If simple consent discriminates, then it is scientifically valid for use in research, irrespective of moral or ethical objections.
Many studies in our review distinguished between consenting and forced acts. We merely compiled the relevant data and examined the value of consent as a predictor of outcomes. It had utility, it did discriminate, and it was therefore scientifically valid to use as a construct. The studies we reviewed generally defined CSA either as a child or adolescent's sexual experience that was unwanted regardless of partner's age, or as wanted or unwanted experiences with someone older typically, at least 5 years older. We merely contrasted study effects from these two groups to examine the value of "consent" as a predictor of outcomes. This analysis clearly demonstrated the utility of distinguishing unwanted from wanted (i.e., consenting) experiences in terms of predicting outcome."
In their November-6th statement, the authors say (p. 14): " It should also be made clear that when Congress, the Leadership Council, the Family Research Council, or even the APA is talking about "children" in the context of sexual relations with adults, they are not using biological definitions of childhood, but instead are referring to minors under the age of consent, which is generally from 16 to 18 in the U.S. Thus, they are talking not only about prepubescent children, but also adolescents. It is thus informative to review what the APA has had to say in the past about adolescents' ability to provide informed consent in a different context. In an October, 1989 amicus brief to the U.S. Supreme Court, the APA argued, based on a review of the developmental literature, that pregnant girls do not need parental consent to obtain abortions, because they are capable, in an informed consent sense, to decide for themselves. They wrote:
'Psychological theory and research about cognitive, social and moral development strongly supports the conclusion that most adolescents are competent to make informed decisions about important life situations. . . . In fact, by middle adolescence (age 14-15) young people develop abilities similar to adults in reasoning about moral dilemmas, understanding social rules and laws, and reasoning about interpersonal relationships and interpersonal problems. . . . By middle adolescence most young people develop an adult-like identity and understanding of self. . . . Thus, by age 14 most adolescents have developed adult-like intellectual and social capacities including specific abilities outlined in the law as necessary for understanding treatment alternatives, considering risks and benefits, and giving legally competent consent. . . . [Additionally,] there are some 11-to-13-year-olds who possess adult-like capabilities in these areas.'
In view of these conclusions, which are based on the developmental literature, it seems inconsistent to reject even simple consent as a moderating variable in a rigorously peer-reviewed article, given that many of the CSA episodes analyzed involved adolescents. In short, the scientific data demonstrate the utility of consent, in the sense of simple consent or willingness, as a moderating variable. Thus, simple consent is a valid scientific construct for predicting and understanding the outcomes associated with CSA experiences."
Several opponents have said that the Rind et al. conclusions were too different from the professional literature to give them any credibility. ‘One-Voice/ACAA’, an alliance of adult survivors of sexual abuse, wrote: "Given the fact that hundreds of studies reveal the serious and long-term consequences of CSA, it is irresponsible to claim that sexual abuse rarely results in harm."
The college data had never been systematically reviewed before. Therefore it is not accurate to conclude - as some of its critics imply - that Rind et al. came to different conclusions from other researchers; it is simply that no other researchers have ever done what Rind et al. have done. Prior to now, there has been no systematic attempt to compare the various college studies with one another in a meaningful way. Therefore Rind et al. cannot be said to have peers who disagree with them; in fact, they have no peers at all; they are pioneers.
One-Voice/ACAA speaks of "hundreds of studies" without the title of a single study: which studies? The Rind et al. team has reviewed all studies they could find.
The authors suggest that researchers should choose more neutral terms for ‘child sexual abuse’ because ‘abuse’ implies harm and there is not always harm. With neutral terms, the real harm can get the attention it needs. Opponents, however, suggest that using neutral terms implies condoning immoral acts.
The authors said in their May-12 statement: "This suggestion was directed to scientific research definitions only, not social or legal ones. The research concern is that although all such sexual contacts may meet legal and social definitions of abuse, the data suggested that such broad definitions may fail to account for how the content of the experience affects psychological outcomes, and that failures to draw distinctions among categories of CSA may compromise the ability to predict effects of these experiences. In other words, all abuse is not equally harmful and suggesting that it is trivializes the most serious abuse.
"We state in the paper that defining CSA based on moral and legal criteria is appropriate for legal restrictions, but may be invalid in scientific inquiry. The term "adult-child sex" already occurs frequently in the child abuse literature, used interchangeably with "child sex abuse." Our paper suggests only that these terms be used more selectively in psychological research - not in media discussions or in legislative discussions."
Opponents suggest that, if NABMLA exclaims "Good news!", it is in fact bad news because pedophiles might feel free to ‘molest our children’.
The conclusion that there is less harm than has always been supposed and that children are more resilient than was thought, is a message of hope. The Rind study is nothing more than another confirmation that children are resilient. There are many studies showing that a percentage of children are able to endure horrific experiences in childhood and yet go on to lead normal healthy lives without apparent damage. People accept such a conclusion when the experiences concern things like deaths of parent or siblings, car accidents, fires, war, or natural disasters. They seem unwilling to accept the same result showing up in this particular circumstance.
Carol Tavris wrote in the L.A. Times (July 10): "Perhaps the researchers' most inflammatory finding, however, was that not all experiences of child-adult sexual contact have equally emotional consequences nor can they be lumped together as "abuse." Being molested at the age of 5 is not comparable to choosing to have sex at 15. Indeed, the researchers found that two-thirds of males who, as children or teenagers, had had sexual experiences with adults did not react negatively.
"Shouldn't this be good news? Shouldn't we be glad to know which experiences are in fact traumatic for children, and which are not upsetting to them? Shouldn't we be pleased to get more evidence of the heartening resilience of children? And "more" evidence it is, for abundant research now shows that most people, over time, cope successfully with adversity - even war. Many not only survive, but find meaning and strength in the experience, discovering psychological resources they did not know they had."
If there is harm – and there is harm in some cases – than it’s better to know which cases are the most harmful. Those are the cases in which the child suffers from a bad family environment, which has far more influence than the sexual experiences. Well, this is ‘bad news’ for organizations that want to keep and protect ‘Family Values’
The authors said in their May-12 statement: "In response to the suggestion that reporting that CSA may be less damaging than previously thought condones abuse, consider what psychologist John Currie wrote in the Atlanta Journal and Constitution (April 8): "To excoriate the APA for publishing the study is a bit like scolding the American Cancer Society for reporting that the long-term effects of a form of cancer are less catastrophic than had previously been thought." This analogy is not just hypothetical. In a study published earlier this year in the journal Pediatrics by R. Noll and colleagues reported in The New York Times, Jan. 19, 1999, (D 12 by Jane Brody), a comparison between nearly all children with cancer in the Cincinnati area and normal controls found that the cancer patients were as well adjusted socially, emotionally, and psychologically as the controls. The authors commented that their study was an improvement over previous research because it examined children in an everyday setting rather than a clinical one. A response to this study analogous to the criticism of our own would be to claim that the authors and the AMA are promoting cancer because it's less harmful psychologically than previously thought."
"In fact, if adverse childhood events are found to be less psychologically harmful than previously thought, or in some cases not measurably harmful at all, researchers have an ethical duty to report this. In the case of CSA, this finding has some positive implications: victims do not have to believe that they are "damaged goods" and will inevitably suffer personality disorders and other psychopathology, and clinicians may have solid grounds for providing reassurance and hope to those who have had such experiences. Ignoring such data may bring harm to those who have had such experiences by perpetuating feelings of being ‘damaged.’ "
Otherwise, in my personal opinion, the "good news!" cry of NAMBLA may not be interpreted as a green light for sexual acts with children; there is less harm than we had thought, but still there is harm in some cases.
Since Laura Schlesinger has said it, many opponents followed her attack on the authors. Those who brought the message, the ‘bad news’, had to fear for their name, their job and life. The authors could have traveled all over the world to promote pedophilia, could have contact with Dutch pedophilia advocates. "The authors write that pedophilia is fine… as long [as] it is enjoyed", to give only one quote from Rep. Joseph R. Pitts, R-Pa, in a press release.
R. Pitt could not give any quotation from the study that supports his remarkable reading (said Jonathan Rauch in National Journal Aug. 6).
The authors said in their May-12 statement: "[…] critics have implied that [our] conclusions condone sexual abuse. In fact, in our article, we clearly state that our review of the research literature does not condone CSA, and changes nothing with regard to moral or legal views of abuse. We wrote that "lack of harmfulness does not imply lack of wrongfulness," that moral and legal codes of society need not be (and often are not) based on findings of psychological harmfulness, and that "the findings of the current review do not imply that moral or legal definitions of or views on behavior, currently classified as CSA, should be abandoned or even altered" (p. 47).
If one does not like the conclusions, let one not attack the authors. One should not give non-existent ‘quotations’. If one will be ethically and morally correct, one should follow the commandment ‘Thou shalt not bear false witness against thy neighbor’.
Dr. Tana Dineen wrote in The Ottawa Citizen, May 11,: "There are two moral issues here. The one that Dr. Laura raises, I support. Pedophilia is wrong.
"The second issue is that of professional honesty. If psychologists truly do not know, are they not obliged to stop telling clients and the courts that child sexual abuse is inevitably harmful. […] The question is: Are they going to admit their lack of knowledge or are they going to quietly cover-up their earlier errors?"
Professional honesty is not only a right of the Rind team, but an ethical obligation for all professionals.
On July 12, the Congress of the USA condemned the meta-analysis in a vote of 355 to zero. To be precise, it especially "condemns and denounces all suggestions in the article `A Meta-Analytic Examination [etc…] that indicate that sexual relationships between adults and 'willing' children are less harmful than believed and might be positive for '`willing' children."
What’s happening here?: politicians condemn and denounce the conclusions of scientific research on moral grounds, after a right wing campaign in the media. If everybody should act morally correct, what should be each one’s role and limits?
Science can tell us that it is healthy to eat meat; human beings can decide on moral grounds to not eat meat. Science can tell us that it is dangerous to drink alcohol; human beings can decide to permit each other to drink alcohol. Science can tell us that it is not dangerous to use cannabis in small portions now and then; human beings can, with draconian punishment, forbid each other to use or possess a tiny portion. Science can give the facts and has the right and the obligation to do this; human beings should hear and read these facts and should discuss honestly the moral consequences. Politicians have the obligation to lead this discussion and to take the decisions as far as necessary.
This discussion, about moral matters, is a different kind of discussion; it differs from the discussion about the facts in every aspect of the discourse. It’s another kind of discourse, as Habermas showed us. The U.S. Congress has interchanged both kinds of discourses.
If politicians with their power (supposedly without reading or understanding the study), decide to condemn and denounce the facts, found in careful scientific research, it’s the end of science, but also the end of a correct discussion about morality. Since most of the media did not read the article at all and gave non-existing ‘quotations’, the public is not well informed and cannot reasonably discuss the moral implications.
Everybody has to accept the conclusions from careful scientific research, until further research gives other conclusions. The FRC wrote: "If psychology finds no harm in something considered morally wrong, we believe they are not looking carefully enough." This is the essence of what passes for respectful criticism of Rind et al. At least, it is not a personal attack. It is, however an attack on the very idea of science. Think what this means: Social scientists would be sent back to the drawing board, until their facts agree with popular prejudices.
Once upon a time, Galileo discovered some facts about the earth and the sun. The Pope refused to accept the facts and with all his power – the power of the Inquisition and the stake – he condemned the scientist. Some centuries afterwards, the Church accepted the facts and gave Galileo his due long after his death.