[Page 715]

"A Meta-Analytic Examination of Assumed Properties of Child Sexual Abuse Using College Samples," published in the July 1998 issue of Psychological Bulletin, resulted in an unprecedented amount of media and political attention. The study' s authors, Rind, Tromovitch, and Bauserman, argued that the media has promoted the idea that child sexual abuse (CSA) "produces intensely negative effects for all of its victims" (p. 22) and that, as a consequence, many professionals assume that sex between adults and children invariably causes "intense" and "pervasive" harm in both genders (p. 22). To test these assumptions the authors analyzed 59 studies of college students. Their article's main findings can be summarized as follows:

Rind et al. (1998) concluded that basic assumptions about CSA were not supported and that the harm associated with sexual relationships between adults and children, especially those involving boys, has been exaggerated - a viewpoint entirely consistent with their prior writings in this area

(e.g., Bauserman, 1989, 1990; Bauserman & Rind, 1997; Rind, 1995a, 1995b; Rind & Bauserman, 1993; Rind & Tromovitch, 1997).

They went on to suggest that when labeling events that have "heretofore been defined socio-legally as CSA," a more "scientific" approach would be to focus on the young person's perception of the experience:

A willing encounter with positive reactions would be labeled simply adult-child sex, a value-neutral term. If a young person felt that he or she did not freely participate in the encounter and if he or she

[Page 716]

experienced negative reactions to it, then child sexual abuse, a term that implies harm to the individual, would be valid. (p. 46)

Rind et al.'s (1998) article provoked enormous social controversy and debate. It was celebrated by pedophile advocates (e.g., Riegel, 2000), drew mixed reviews from psychologists (e.g., Byrd, Scharman, & Lauritsen, 1999; Tavris, 1999); and sparked outrage in the popular media (e.g., Saunders, 1999) and in political circles (see Duin, 1999).

In the end, the study's conclusions were formally (and un--artimously) denounced by both the U.S. House of Representatives (H. R. Cong. Res. 107) and the U.S. Senate. It is believed to be the first time in U.S. history that a scientific study was formally repudiated by a legislative body.

 Despite the controversy surrounding it, the article by Rind et al. has been referenced in legal proceedings. For example, in State of Arizona v. Steward (1999) a convicted child molester used the article to argue for leniency, saying that research shows that children are rarely harmed by sexual molestation. In deposition testimony, a psychologist working as an expert witness for a pedophilic priest relied on the study to form the opinion that a victim's psychological injuries were not due to sexual abuse, because the current scientific literature does not support the existence of a relationship between CSA and maladjustment in the population at large (Brainerd, 1999). The article by Rind et al. has also been cited as evidence that the legal age of consent for sex should be lowered (Graupner, 1999, see Dallam, in press, for a more complete review).

In the current review, we examine the data and methods that Rind et al. (1998) used to form their conclusions. Please note that the purpose of our article is not to argue that all types of sexual abuse do in fact cause pervasive and intense harm in all victims. Indeed, it is well recognized in the empirical literature that the aftereffects of CSA are extremely varied and that a significant percentage of abused children remain a-symptomatic

(Beitchman, Zucker, Hood, DaCosta, & Akman, 1991; Beitchman et al.1992; Browne & Finkelhor, 1986; Kendall-Tackett, Williams, & Finkelhor, 1993).

Instead, because of their important implications for public policy, we seek to examine the validity and generalizability of Rind et al.'s results.

The purpose of meta-analysis is to combine information from independent studies that address a similar question to provide more accurate estimates of the effects being measured. The potential for meta-analysis to add to research knowledge is great, provided that the meta-analysis meets high standards and well-designed studies are available on the subject matter.

Rind et al. (1998) suggested that quantitative literature reviews can avoid "subjectivity and imprecision by using meta-analysis" (p. 24). Although this is certainly the goal, meta-analyses are essentially retrospective analyses of previously collected data. Thus, problems in design or reporting may be uncritically carried into the analysis and influence its results. In other words, the flaws and biases associated with individual studies do not necessarily cancel each other out when studies are combined meta-analytically (Matt & Navarro, 1997). Consequently, it has been recommended that readers look carefully at the studies that were included in a meta-analysis, and require evidence of thoughtful and deliberate conduct from any meta-analytic study before accepting its conclusions (LaValley, 1997).

For our review, we retrieved and examined all published studies analyzed by Rind et al. (1998), along with abstracts of the unpublished dissertations. If a dissertation's abstract did not provide adequate information, a copy of the complete manuscript was obtained. Using guidelines provided by Rosenthal (1994, 1995), we examined criteria for inclusion, study quality, independence, study characteristics, dependent measures, and effect size estimates. We did not, however, attempt to completely replicate the meta-analysis; rather, we looked carefully at how it was conducted and the strength of the evidence used to support Rind et al. '  major findings.