Construct Validity

Content of this article Tables References

Vorige Start Volgende

[Page 751 continued]

Although Dallam et al. (2001) in particular, and Ondersma et al. (2001) as well, vigorously challenged our analyses, coding, and other methods, we believe that much of the reaction to our meta-analysis stemmed from other issues. In particular, we believe that our recommendation regarding more cautious use of the term sexual abuse and our referral to some CSA experiences as consenting triggered the strongest reactions. We now turn to these issues.

A primary thesis of Ondersma et al. (2001) is that we misused science by means of "extra-scientific" (p. 712) considerations, especially when suggesting that professionals use more value-neutral language when discussing some CSA experiences. We feel it is important to discuss how this section of our review came about and its scientific justification in terms of construct validity .

In our original drafts submitted to Psychological Bulletin, we did not make any recommendations concerning use of value-neutral terms. We did so in the final draft in response to comments by reviewers and the action editor, who noted that we had found that broad definitions of CSA resulted in poor predictive validity and, by omitting the context of a sexual contact, might fail to adequately capture the "essence" of "abuse" (K. J. Sher, personal communication. May 14. 1997).

In short, the goal was to improve construct validity, such that abuse would be predictive of actual or probable negative outcomes. The section of our Discussion labeled "Child Sexual Abuse as a Construct Reconsidered" and the recommendations offered in our Conclusions section represented our ideas as to how best identify and study those adult-adolescent and adult-child sexual contacts most likely to be associated with harm.

We never argued that the term child sexual abuse should not be used, rather its use should be restricted to experiences (e.g., those involving coercion or force or involving negative reactions) that are more predictive of negative outcomes. Furthermore, we specified that we were concerned with research definitions of CSA, not social and legal definitions, and specifically noted that whether an experience caused harm was not the same as whether it was wrongful. These recommendations were concerned with improving the predictive validity - and in turn, the construct validity - of CSA.

Clearly, alternative approaches are possible. For example, Brant and Tisza (1977) proposed labeling some CSA experiences sexual misuse rather than abuse, to distinguish experiences not associated with harm from those that are. However, we felt it more useful to avoid terms that carried any implications or assumptions of psychological or other harm unless specifically justified by the context.

Ondersma et al. (2001 ), and other critics, implied that we were extreme in suggesting that abuse might not be an appropriate label in some cases.

However, many sexologists have argued that uncritical use of victimological language in CSA research creates problems in scientific validity

(e.g., Green, 1992; Kilpatrick, 1987; Money & Weinrich, 1983; J. A. Nelson, 1989; Okami, 1990; Sandfort, 1992; West, 1998).

Some authors of the college studies argued this as well

(e.g., Fishman, 1991; Fromuth & Burkhart. 1987; Long & Jackson, 1993; West. 1998).

For example, West (1998). a criminologist and sexologist from Cambridge University and coauthor of two of the college studies in our review, argued that professional use of terms such as abuse, perpetrator, victim. and survivor has incorrectly reinforced the idea that any kind of sexual incident with a child is likely to cause great and lasting harm, stating that this usage has introduced a moral tone "alien to scientific inquiry" (p. 539).

After qualitatively reviewing the CSA literature, Green ( 1992), a psychiatrist, lawyer, and editor of the Archives of Sexual Behavior, reached similar conclusions to ours. He stated the following;

Ultimately, scientists, if no one else, must be objective in their approach to this emotional issue. Judgmental terminology regarding intergenerational sexuality is more dramatic than that in the earlier psychiatric literature on homosexuality. There, patients were labeled perverts and psychopaths. Here, the experience is always abuse. the children are invariably victims, the adults are perpetrators, and those who later report childhood sexual experiences are, without apology to victims of the Nazi Holocaust, survivors. (p. 175)

Thus, our recommendations were in line with both scientific principles of validity and sexological precedent. Emotionally loaded victimological terminology carries clear and strong implications about the nature and effects of the experiences it describes. It is reasonable to question whether such implications are always appropriate.

The further one moves away from clinical and psychiatric reports, the clearer it becomes that neutral rather than victimological language is the norm. In Laumann et al.'s (1994) national study, age-discrepant contact sex was simply labeled sexual touching. Many nonclinical studies that are based on convenience samples of boys' sexual experiences with older persons have avoided abuse terminology unless abuse clearly applied to individual cases (see Bauserman & Rind, 1997).

"Abuse" terminology is almost completely absent when moving beyond psychology into anthropology, zoology, and history. Ford and Beach (1951) explicitly stated that "moral evaluations form no part of this book" (p. 14), at a time when many of the forms of sex they analyzed were taboo and/or illegal in the U.S. Hundreds of anthropological and zoological publications since Ford and Beach have avoided such terminology (e.g., Greenberg, 1988; Gregersen, 1983; Vasey, 1995). Because these researchers have not been concerned with moralizing, treating, preventing, and curing, they have been able to approach these phenomena as scientists attempting to describe and

[Page 752]

understand, and they frequently have discussed functions such behaviors appear to serve - something that is not possible if one is constrained by a disease model. The same applies to historical research, in which scholars discussing pederasty in ancient Greece or Rome (e.g., Cantarella, 1992; Percy, 1996) or early modern Japan (e.g., Schalow, 1989) almost invariably use neutral language.

Ondersma et al. ' s (2001) claim that value-neutral language is a threat to the moral fabric because "small but vigorous" (p. 712) minorities will exploit it is an example of the "argument from adverse consequences fallacy" (Lilienfeld, in press; Sagan, 1995). This is the error of evaluating the validity of an argument by considering its potential negative consequences.

Use of neutral language is part of scholarship, even that which deals with CSA. Combined with selective use of terms such as abuse, it enhances predictive validity. Value-laden language, we argue, has the effect of framing all aspects of the issue in a particular moral and political viewpoint. In turn, this obscures the empirical questions of the presence and degree of harm.

Content of this article Tables References

Vorige Start Volgende