2000 adolescents, more than 80 percent of abuse victims did not report the abuse to anyone (Edgardh and Ormstad, 2000). All self-reporting is vulnerable to bias, with particular concern for reports elicited years after events. Some studies have suggested that recall of childhood abuse varies with psychological adjustment (e.g., Falsetti and Resnick, 2000), while others have not found an association (Fergusson et al., 2000a; Robins et al., 1985). Examining this issue in a population-based, longitudinal, prospective study with repeated measures of child abuse self-reports, Fergusson and colleagues (2000a) discovered about a 50 percent rate of forgetting and/or not reporting documented abuse during assessments. They found that lack of recall did not vary with psychiatric diagnosis or suicidality. Other available data on reporting bias of childhood abuse also consistently indicates that abuse is significantly under-reported, with, depending on measures used, 40–60 percent lack of recall for documented cases of maltreatment (Fergusson et al., 2000a; Widom and Shepard, 1996; Widom, 1997; Williams, 1994). The reasons are complex, including forgetting (usually if the victim was less than 5 years old), stigma and embarrassment, relationship to the perpetrator, nature of the abusive or traumatic incident, and sensitivity of the survey or interview measures (see for example Kessler, 2000; Williams, 1994).
Another concern about self-reports, particularly with sexual abuse, regards repression of memories as a means of self-protection, with later recovery in adulthood. Repression could lead to either false positive or false negative reporting, but the evidence for repression appears to be controversial (Berliner and Williams, 1994; Loftus et al., 1998). There are no data to indicate what percentage of “recovered memories” are inaccurate, but data indicate 47–95 percent of recovered memories of non-bizarre child abuse are confirmed, and only 1–3 percent of bizarre abuse memories are confirmed (Bowman, 1996a; Bowman, 1996b). A recent study demonstrated that 74 percent of both always recalled and recovered memories could be confirmed from a legal point of view (Dahlenberg, 1996). The aforementioned analysis of longitudinal data by Fergusson’s team (Fergusson et al., 2000a) further suggests that forgetting and later recall of childhood abuse represents a common phenomenon not associated with psychopathology, though they could not distinguish between active repression and simple forgetting. These investigators caution that recall bias obscures true prevalence rates of child maltreatment, though it does not, it appears, significantly alter estimates of relative risk of child abuse for subsequent psychological disorders.
A second limitation in the current research on childhood abuse is the use of inconsistent and imprecise definitions of maltreatment (NRC, 1993). Definitions may vary among mandated reporters, both within and across agencies, localities, and states, thereby affecting official reporting statis-