Some Methodological Problems in Estimating Incidence and Prevalence in Child Sexual Abuse Research.

Many areas of social research are riddled with problems of methodology. Child sexual abuse is no exception. In the last 30 years much of the research into child sexual abuse has emanated from the United States, and to a lesser extent from Europe and Australia, producing an increase in both the quantity and quality of its research. The number of countries where child sexual abuse has been examined has been extended significantly (see Finkelhor, 1994). The variety of theoretical approaches has also widened to include sociological, psychological, psychoanalytical, historical, feminist and others. Consequently, researchers now have a much greater understanding of the topic from a range of theoretical perspectives. Due to the increase in child sexual abuse literature, numerous reviews have appeared that attempt to synthesize the studies. However, as Finkelhor (1994) noted when comparing data from 21 countries including the USA, Canada and Australia, "few comparisons among countries are possible because of methodological and definitional difficulties" (p.409). Such difficulties extend not only to comparing countries, but to the whole area of child abuse research. Consequently, several reviews in the area focus on highlighting these methodological difficulties (e.g., Briere, 1992; Gorey & Leslie, 1997; Roosa, Reyes, Reinholtz, & Angelini, 1998; Wyatt & Peters, 1986; Wynkoop, Capps, & Priest, 1995), with a more recent focus being on difficulties in assessing the child sexual abuse-adult adjustment relationship (e.g., Briere, 1992; Rind & Tromovitch, 1997; Roosa et al., 1998). However, methodological difficulties exist at a more basic level, with problems in simply identifying how many people have suffered from child sexual abuse. Therefore, the present article focuses on providing an up-to-date and comprehensive review of the difficulties faced when trying to obtain an accurate estimate of the number of people who have been abused. In doing so, the article's uniqueness lies in its attempt to unpack some of the issues which underlie the disparate conclusions drawn in a variety of child sexual abuse research, particularly in incidence and prevalence studies.

THE DIMENSION OF THE PROBLEM

There has been substantial debate for a number of decades on the actual number of children being sexually abused. Researchers have had difficulty in determining exactly how many children are being sexually abused each year, whether in the USA, Canada, Europe, Australia, or other western countries. Efforts have been made to collect data from children referred to police, social services, and hospitals. Such incidence studies normally estimate the number of new cases occurring in a given period of time (Peters, Wyatt & Finkelhor, 1986; Wyatt, & Peters, 1986).

Another source of data is from population studies in which adults are surveyed to provide retrospective evidence of their sexual experiences as children. Such prevalence studies (Leventhal, 1990; Peters et al., 1986) are attempts to estimate the proportion of a population that has been sexually abused during their childhood, and should use national probability sampling to obtain the most representative results. However, prevalence studies appear to be less common than incidence studies.

Incidence Studies

The following overview examines some of the incidence studies that have estimated the number of children currently being sexually abused by adults. Extrapolating from reported incidents of sexual abuse of children in four different American locales, Sarafino (1979) estimated the national incidence of reported and unreported child sexual abuse to be over 336,000 cases per year. Sarafino arrived at this figure by calculating the rates of reported sexual offences per 100,000 children in each of the four locales, and then applying this rate to the national total of 61 million children. This led to an estimated 74,725 cases of child sexual abuse in a one-year period. The rate of unreported cases was calculated by multiplying 74,725 by 3.5 (assuming that the number of unreported cases is at least 3 or 4 times higher than the reported cases as believed by several experts in the field, e.g., De Francis, 1969; Hayman & Lanza, 1971). The number of reported cases was added to the estimated number of unreported cases. Consequently, it was estimated that approximately 336,200 sexual offences are committed against children every year in the United States.

However, there are several limitations to this study. For instance, this study is based on data from a limited number of locales. Moreover, there is little standardization as different states and local communities may use different terms and legal definitions in categorising sex offences. Further, the projections derived from other estimates--for example, the ratio of reported cases to unreported cases--is unreliable, thus making the results questionable.

In a study in Britain, Mrazek, Lynch, and Bentovim (1983) estimated an incidence of about 1,500 cases of child sexual abuse per year, which represents one child in every 6,000. This estimated figure was extrapolated from a survey of 622 professionals (response rate of 39%) who directly dealt with sexually abused children. Since the remaining professionals did not respond to the questionnaire, it is difficult to know whether this is a reasonable estimate.

In a national study in the United States by National Centre on Child Sexual Abuse and Neglect (NCCAN) (1981), of all the cases known to professionals it was estimated that 44,700 children are sexually abused every year. Again, data from this study may be unreliable as it is often found that families from lower socio-economic class with fewer resources would turn to public institutions such as a Department of Social Security for help, whereas families from middle and higher socio-economic class would resort to private means (De Jong, Hervada, & Emmett, 1983; Scherzer & Lala, 1980). Further, physicians often may not report, or advise the family of the victim to report the abuse to the appropriate authorities. For instance, James, Womack, and Stauss's (1978) study of physicians revealed that only one third of physicians urged the family of sexual abuse victims to report the matter, and less than half indicated that they would do so. A recent study in Australia found that physicians had very little knowledge of the experience of sexual abuse in their female patients (Mazza, Dennerstein, & Ryan, 1996). These findings clearly demonstrate that the incidences of child sexual abuse are underreported, and cannot be accurately estimated from recorded cases.

Overreporting of cases of child sexual abuse is also not uncommon, as found in the much publicized Cleveland Inquiry in Britain (see O'Hagan, 1989). It was found that within a 5-month period, 121 children who were diagnosed as victims of child sexual abuse on limited medical examination were removed from their homes. Of these, 98 were returned to their homes at the time of the inquiry as it was found that they appeared not to have been abused. This appears to be an example of a case of overreporting.

In Australia, some attempts were made by Harrold (1987) to estimate a national incidence figure for child abuse, including child sexual abuse, by examining notification records in each state. Harrold found that, similar to the USA, it was not possible to obtain an accurate assessment of the incidence of child sexual abuse or compare the reported rates between states. This was due to differences in the definition of child abuse and administrative procedures for handling cases once the authorities had been notified. For instance, in Queensland, a case is notified and recorded as soon as the initial complaint is made, whereas some other Australian states only notify and record after initial investigations have been made. Since it was methodologically unsound to compare reported rates across states in Australia, Harrold presented some information on incidence figures for child abuse in Queensland only. Based on the data from the Central Register, Harrold concluded finding a dramatic increase in the confirmed cases of child sexual abuse from 60 cases for the period 1981 to 1982 to 184 cases for the 1983 to 1984 period.

Although reported cases of child sexual abuse do not give a good estimation of the magnitude of the problem, they help to keep track of the number of cases which come to the attention of professionals and social agencies. However, many researchers believe that reported cases of child sexual abuse represent only a fraction of its true occurrence. Only the most severe cases are reported to the legal authorities and social agencies. Beneath the visible tip of child sexual abuse is a larger portion of unknown cases (Finkelhor, 1984). Since so many cases are unreported due to the nature of the problem, a much broader population study may determine prevalence. A number of prevalence studies will now be reviewed which represent significant attempts to sample the general population, and illustrate how different methodologies, definitions, and samples influence results.

Prevalence Studies

Since the absolute number of children being sexually abused each year has been almost impossible to ascertain, a number of researchers have attempted to make some estimate of prevalence rates in stratified samples such as married women or college students in retrospective self-reporting (Finkelhor, 1979; Finkelhor, Hotaling, Lewis, & Smith, 1990; Goldman & Goldman 1988a; Goldman & Padayachi, 1997, 2000a, 2000b; Kinsey, Pomeroy, Martin, & Gebhard, 1953; Landis, 1956; Landis et al., 1940). However, in recent studies, a concerted effort has been made to estimate the prevalence using probability samples (Finkelhor, 1984; Russell, 1984). Even so, there does not seem to be agreement on the rate of children being sexually abused.

Some researchers have estimated that over 50% of the female child population will experience some form of sexual abuse before the age of 18 (Russell, 1984; Wyatt, 1985), while others have reported rates of 11% and lower (Fritz, Stall, & Wagner, 1981; Kercher & McShane, 1984; Seidner & Calhoun, 1984). Similarly, while a meta-analytic study by Rind, Tromovitch, and Bauserman (1998) reported mean prevalence rates of 17% and 28% for males and females respectively, the range for males was 3% to 37%, and for females 8% to 71%. Such wide variation in the prevalence rate is due to differences in the definition of child sexual abuse, the type of sample used, design, and measurement techniques.

METHODOLOGICAL PROBLEMS IN PREVALENCE RESEARCH FINDINGS

In order to analyse some methodological problems, a summary of the available results from a selection of prevalence studies of child sexual abuse is presented in Table 1. The studies have been examined in terms of the year of survey, country where the survey was conducted, type of sample, response rate, definitions of sexual abuse and child, rates reported for females and males, and other factors. Table 1 indicates considerable variation in prevalence rates. Prevalence rates for females range from 7% (Siegel, Sorenson, Golding, Burman, & Stein; 1987) to 62% (Wyatt, 1985), and for males from 4% (Siegel et al., 1987) to 30% (Landis, 1956). Several possible factors which may account for the discrepancies in prevalence rates will now be discussed.

[TABULAR DATA 1 NOT REPRODUCIBLE IN ASCII]

USE OF DISPARATE DEFINITIONS OF SEXUAL ABUSE

A common reason for the discrepancies in prevalence rates is the use of disparate definitions. There appear to be four major factors which have to be to considered before sexual experience can be regarded as sexual abuse. These include how sexual contact is defined; whether noncontact sexual events such as exhibitionism, pornography, or sexual invitation is included; whether all wanted and unwanted sexual experiences or just unwanted sexual experiences are included; and whether an age differential is employed between the victim and the perpetrator.

In the literature some researchers exclude peer sexual abuse in their data (e.g., Finkelhor, 1979, 1984; Fromuth, 1983; Goldman & Goldman, 1988a; Goldman & Padayachi, 1997; Seidner & Calhoun, 1984) whereas others include this form of abuse (Russell, 1984; Siegel et al, 1987; Wyatt, 1985). Some use only contact sexual abuse (Bagley & Ramsay, 1986; Briere & Runtz, 1988; Fritz et al., 1981; Kercher & McShane, 1984; Mullen, Roman-Clarkson, & Walton, 1988; Siegel et al., 1987) whereas others use both noncontact and contact abuse (Badgley et al., 1984; Baker & Duncan, 1985; Finkelhor, 1979; 1984; Finkelhor, et al., 1990; Goldman & Goldman, 1988a; Goldman & Padayachi, 1997; Kinsey et al., 1953). A recent study by Roosa et al. (1998) illustrates the difference the definition of child sexual abuse can make to reported rates of abuse. They contrasted five- and six-level definitions of sexual abuse. The five-level definition included the categories of no sexual abuse, contact molestation, coercion, attempted rape, and rape. The six-level definition included the additional category of noncontact molestation, and broadened the definition of contact molestation. Results showed that when the five-level measure of sexual abuse was utilized, 39% of respondents were identified as victims of child sexual abuse, compared to 59% when the six-level measure was used. Also, when women who were victimized by boyfriends and friends were not considered to fit the definition of sexual abuse, the rates of abuse for the five- and six-level measures were 18% and 34% respectively. Consequently, it appears that such research which uses broader definitions, such as the inclusion of both contact and noncontact cases, both unwanted and wanted events, and no age discrepancies between the perpetrator and the victim, report higher prevalence rates.

Definition of a Child

The age demarcation of the victim as a child also appears to affect prevalence rates. As shown in Table 1, some researchers have stated that the victim needs to be younger than 18 years of age, since 18 is often the age of legal consent in the United States (Badgley et al., 1984; Russell, 1984; Wyatt, 1985). In cases where the child was defined as a person under 18, the majority of the investigators reported fairly high rates. Those researchers who use younger age demarcations have reported lower prevalence rates (Fritz et al., 1981; Kercher & McShane, 1984; Mullen et al., 1988; Sedney & Brooks, 1984). Thus, age demarcation seems to affect prevalence rate.

Historical Time When Survey Conducted

Some researchers of child sexual abuse believe that the time when the survey was conducted also affects prevalence rates (Leventhal, 1990). Studies conducted during a period when the problem of sexual abuse is widely publicized may elicit higher rates as adults may be more willing to disclose their experience of sexual abuse. Surveys conducted prior to the 1970s report lower rates compared to more recent ones (e.g., Kinsey et al., 1948, 1953). Respondents prior to the 1970's may have been more reluctant to talk about sexual matters because of social, moral, and religious sanctions (see also Goldman & Goldman, 1982). However, by the mid-1970s, social recognition of the problem of child sexual abuse and the incorporation of federal and/or state laws mandating reporting of such cases to child protective services in many countries may have made it easier for victims to report what had happened to them.

Rate of Child Sexual Abuse increase

It is often believed that there has been a sharp increase in the number of sexual offences committed against children in the later part of the 20th century. Sexual freedom, pornography, and the permissive attitudes since the 1960s, may have contributed to the increase in the prevalence of child sexual abuse. If this were so, then there should be higher prevalence rates among younger cohorts within a given sample or in a college sample. However, a review of the literature reveals conflicting evidence of this.

For example, Russell (1986) divided her sample into two cohorts: 18-36 years of age and 37 years and older, and calculated prevalence rates separately for the two cohorts. She found that the prevalence rate in the younger cohorts (43% for contact sexual abuse and 59% for noncontact and contact sexual abuse combined) was higher than it was among the older group (34% and 50% respectively). Briere (1992) provides a possible explanation for such results, arguing that age-specific socialization may influence a person's willingness to report sexual abuse. That is, individuals who grew up in an earlier era may be more likely to deny sexual abuse because of the greater stigmatization associated with sexual abuse in previous decades. However, this explanation does not address results such as those found by Wyatt (1985), whose study revealed that the younger cohort (aged between 18-26 years) had a slightly lower rate (41% for contact sexual abuse and 60% for contact and noncontact sexual abuse combined) when compared to rates among the older cohort aged between 27-36 years (48% and 64% respectively).

Finkelhor (1979) made a detailed comparison of data collected in three different studies, namely Kinsey and associates (analysed by Gagnon, 1965), Landis (1956), and his own survey of college students undertaken in 1978. He concluded that sexual molestation of girls in the last 30 years has probably stayed the same (p. 134).

However, data from the reported cases of child sexual abuse reveal there are more reports made to the Child Protective Services, the police, other social services, and treatment agencies in the USA than previously (Giarretto, 1982; NCCAN, 1981). There are several reasons for this increase. Finkelhor (1979) believes that the increase in the reports made to the police and the protective services does not necessarily mean that there is an increase in the actual occurrence of child sexual abuse. According to Finkelhor (1979):

   What we are witnessing is a revolution in consciousness, a situation where,
   because of changed mores, professionals are more sensitive to identifying
   instances of sexual abuse and victims and their families are more willing
   than before to seek help. (p. 132)

Moreover, in light of the increase in population growth in many countries, there is no doubt that over time the volume of child sexual abuse has increased. However, there are no historical prevalence data available to determine whether proportionately more persons are abused at present than in the past. Once again methodological problems are evident.

Sexual Abuse of Boys

There are fewer methodological problems associated with gender differences since most research into child sexual abuse distinguishes between male and female victims. A review of clinical and empirical studies show lower rates of male victims compared to female victims of child sexual abuse. Almost all researchers in Table 1 reported the rate of sexual abuse of boys to be under 16% (the exception to this is Landis, 1956, who reported a prevalence rate of 30%). Many. researchers believe that the rates of sexual abuse of boys and the number of cases that come to the attention of clinicians and social agencies is relatively small (Finkelhor, 1981, 1984; Pierce & Pierce, 1985).

However, there is also a growing recognition that the sexual abuse of boys is underreported (Briggs & Hawkins, 1995; Cermak & Molidor, 1996; Violato & Genuis, 1993; Watkins & Bentovim, 1992). Several explanations for this occurrence have been suggested. One such explanation is that socialization of males is different to that of females. Boys are socialized to be strong, to be self-contained, and to not reveal their emotions. The fear of exposing their vulnerability may be one reason why males do not report their sexual experiences to an adult. However, an interesting change is currently occurring in Australia, where the New South Wales Royal Commission into Police is receiving evidence from tens of adult males, sexually abused as boys, who are coming forward to reveal the perpetrators and the nature of their abuse. Such a voluntary sample helps address methodological difficulties associated with gender differences.

Another reason why male victims seldom disclose their experiences of sexual abuse is because of the stigma attached to being sexually abused by another male, and the fear of being labelled "queer" or "homosexual" (Black & De Blassie, 1993). This is evident with many studies showing a preponderance of male perpetrators (e.g., Finkelhor, 1979; Goldman & Goldman, 1988b; Goldman & Padayachi, 1997; Russell, 1984). Finally, the low incidence figures for males in the research literature and from legal and social agencies may be attributed to different definitions employed. These factors may help clarify low rates reported for boys.

PERPETRATORS

Women Perpetrators

In the methodology in studies of child sexual abuse, the gender of the perpetrator is frequently included. Until recently, it was assumed that women rarely sexually abused children, but studies now acknowledge the existence of female perpetrators (Goldman & Goldman, 1988a; Goldman & Padayachi, 1997). A review of all the available data in the United States on this subject was conducted by Russell and Finkelhor (1984). They found that in the reported cases of child sexual abuse, less than 24% of female victims and less than 13% of male victims were sexually abused by female perpetrators.

Similarly, Russell and Finkelhor's (1984) review of the self-report studies reveals that, of the male victims of child sexual abuse, 27% or less were abused by women, with one exception. The exception is the figure reported by Fritz et al. (1981) that 60% of the boys in their study were abused by women. Russell and Finkelhor believe that this figure could have resulted from an unusual sample or an error in tabulation.

For the female victims in the self-reported cases, less than 10% of them were abused by women, again with one exception reported by Bell and Weinberg (1981). Bell and Weinberg used homosexuals, lesbians, and heterosexuals as their sample. Their data revealed that 22% of the lesbians reported being sexually abused by other women, which is considerably higher compared to other studies with heterosexual samples. For instance, Finkelhor (1979) reported 6%, Finkelhor (1984) reported 0%, Fritz et al. (1981) reported 10%, Fromouth (1983.) reported 5%, and Russell (1984) reported 4%. The high percentage of lesbians being abused by female perpetrators reported in Bell and Weinberg's study cannot be generalized to the wider population. Lesbians may have a different sexual configuration compared to heterosexual women which may make them more vulnerable to sexual experiences with other women. Lesbians' early sexual experiences with other women may have influenced them to become lesbians. Whatever the explanation, Bell and Weinberg's findings do not confirm that a high percentage of children are being sexually abused by women.

Another study that found a high incidence of female perpetrators was conducted by Finkelhor, Williams, Burns, and Kalinowski (1988). This study investigated levels of sexual abuse in a day care setting and found that 40% of the abusers were females. The unusually high incidence of women as perpetrators may be explained by the low presence of male workers in day care centres.

Overall, such results are contrary to the long-held assumption that women do not sexually abuse children. Finkelhor and Russell (1984) estimate that females are the perpetrators in approximately 5% of girl victims and 20% of boy victims and there is considerable evidence in the literature that suggest that these figures are reasonably accurate estimates (Finkelhor, 1979; Fromouth, 1983; Goldman & Goldman, 1988b; Goldman & Padayachi, 1997; Russell, 1984).

Preponderance of Male Perpetrators

Whatever the methodology employed, studies consistently show that the overwhelming majority of perpetrators are males (NCCAN, 1981). Several researchers have explored explanations for this (Finkelhor & Russell, 1984; Gilgun, 1991; Goldman & Goldman, 1988a; Goldman & Padayachi, 1997; Herman, 1981a, 1981b; Padayachi, 1994). One theory is that sex role socialization is a contributing factor in the development of sexually abusive behaviour toward children. Men are socialized to be aggressive, and link sexual conquests and aggressiveness to masculinity (Gilgun, 1991). Additionally, men who have been abused when they were children tend to act out in sexual or sexually violent ways. In contrast, women are exposed to nurturing role models and tend to reveal emotions with others, whereas men seldom do. Gilgun reports that all of the 48 subjects in her study were maltreated as children. Of the 36 male subjects, 15 were perpetrators of child sexual abuse and 12 were convicted of rape, attempted murder, armed robbery, or murder. None of the convicted felons reported that they disclosed their emotions to anyone. Men are discouraged from doing so, whereas women are encouraged to confide or solve their problems through talking. Thus, men may tend to act out their frustrations and unresolved emotions both physically and sexually.

Choice of Sample

Methodological problems are also evident in child sexual abuse research in the choice of sample. There appear to be three major methods employed in the literature for selecting a sample for prevalence studies of child sexual abuse. These include random sampling, nonprobability sampling of college students from a particular class in college, and requesting volunteers from the population.

Prevalence rates appear to be affected by the choice of samples. For example, in studies using student samples, Finkelhor (1979) found a prevalence rate of 19%, Fritz et al. (1981) a rate of 8%, and Fromuth (1983) a rate of 22%. In contrast, Badgley et al. (1984), Russell (1984), and Wyatt (1985) used community samples and found abuse prevalence rates to be 54%, 34%, and 62% respectively. (See Table 1 for details.) One possible reason for fewer victims being found in a college sample is because they may be a psychologically healthier group. A number of studies have shown that girls from families of lower educational level are at greater risk for child sexual abuse (Finkelhor, 1980), and their lower educational achievement may not have met the college academic entry requirements. Studies using community samples such as those by Badgley et al. (1984), Russell (1984), and Wyatt (1985) may report higher prevalence rates compared to studies using college samples because they include these more poorly educated people.

However, there appear to be no clear trends as results from Rind et al.'s (1998) meta-analytic study indicated that the overall prevalence of child sexual abuse was not lower in the college sample when compared to national samples. Furthermore, other studies using community samples have reported low prevalence rates. For instance, Siegel et al. (1987) reported that of the total female sample of 1,645, only 7% were sexually abused. Similarly, other community-based population studies report fairly low rates of child sexual abuse (e.g., Baker & Duncan, 1985; Finkelhor, 1984; Kercher & McShane, 1984; Mullen et al., 1988). This suggests that other factors and possible unwillingness to divulge such information, apart from sample choice, has some effects on prevalence rates.

Differences in Data Collection

Another reason for the variation in the estimates of child sexual abuse may be methodological problems due to the different modes of data collection. Two aspects of data collection which appear to account for the discrepancy in the findings are data collection procedures and the structuring of questions.

Table 2 shows the different modes of data collection used in the prevalence studies in North America in the last two decades. Studies from North America only are included in this table for the purpose of comparison and to analyse how prevalence rates differ in terms of modes of data collection. The three modes of data collection employed by researchers are self-administered questionnaires, face-to-face interviews, and telephone interviews.

Table 2. Prevalence Rates for Women by Mode of Data Collection

Researcher(s)                           Date      Rate %

Self-administered questionnaire
  Badgley et al.                        1984        34
  Briere & Runtz                        1988        15
  Finkelhor                             1979        19
  Finkelhor                             1984        15
  Fritz et al.                          1981         8
  Fromuth                               1983        22
  Kercher & McShane                     1984        11
  Miller                                1976        14
  Sedney & Brooks                       1989        16
  Seidner & Calhoun                     1984        11

Face-to-face interview
  Bagley & Ramsay                       1986        22
  Russell                               1984(b)     54
  Siegel et al.                         1987         7
  Wyatt                                 1985        62

Telephone interview
  Finkelhor                             1990        27
  Keckley Marketing Research            1983        11
  Lewis                                 1985        27

Note. Data collected in North America between 1970 and 1990.

In Table 2, there appears to be some relationship between the mode of data collection and the prevalence rates of child sexual abuse. Of the 10 studies which used self-administered questionnaires, only one reported a prevalence rate above 30% and the rest reported rates within the range of 8% and 22%. The general consistency of rates may be explained as studies which utilized self-administered questionnaires used almost entirely student samples (except Finkelhor, 1984) and used different versions of Finkelhor's (1979) questionnaire. In contrast; this homogeneity is absent in the four studies which utilized face-to-face interviews. That is, probability samples and varying methodologies were used. One study which used face-to-face interviews reported very low prevalence of fewer than 8% (Siegel et al., 1987), and two reported fairly high rates of over 50% (Russell, 1984; Wyatt, 1985). Of the three studies using telephone interviews, two reported fairly low rates of below 14%, and two reported 27%.

It should be noted that all three of these data collection techniques are prone to difficulties inherent in retrospective reporting. There is always a question of whether what is being reported is, in fact, what happened. Briere (1992) suggests that the passing of time and also repression of memories of the trauma can lead to inaccurate results.

Of the three modes of data collection used, however, face-to-face interviews appear the most likely to be able to reduce the effect of such difficulties. In face-to-face interviews, trained caseworkers are often used, and in most interviews the race and ethnicity of the interviewer and respondent were matched (e.g., Russell, 1984; Wyatt, 1985). Therefore, the caseworkers may have built up a rapport with the subject, enhancing the subject's motivation to respond more freely (Bradburn, 1983). Moreover, the interviewer may convey a nonjudgmental, caring, and empathetic approach, thus encouraging the respondents to disclose more information about their sexual abuse. The interviewer may also have attended to cues that showed the respondent was uncomfortable, or may have assisted if the respondent was misinterpreting a question. These factors show that face-to-face interviews appear to have an advantage over self-administered questionnaires or telephone interview methods, and may have considerable effects on the prevalence rates.

Type and Number of Questions

The type and number of questions asked to elicit data on child sexual abuse also make a crucial difference to prevalence rates. Higher rates were reported in studies that allowed for maximum probing of the respondents. For instance, those studies which used only one screening question reported fairly low rates compared to those which used more than one screening question (see Table 3).

Table 3. Prevalence Rates for Women by Number of Screening Questions

Researcher(s)            Date      No. of Questions    Rate %

Bagley & Ramsay          1986             1              22
Keckley Market Report    1983             1              11
Kecher & McShane         1984             1              11
Miller                   1976             1              14
Siegel et al.            1987             1               7
Seidner & Calhoun        1984             2              11
Finkelhor                1984             3              15
Badgely et al.           1984             4              34
Finkelhor                1979             4              19
Finkelhor                1990             4              27
Lewis                    1985             4              27
Wyatt                    1985             8              62
Russell                  1984(b)          14             54

Note. Data collected in North America between 1970 and 1990.

Studies that used more than four screening questions reported very high rates of child sexual abuse (e.g., Russell, 1984; Wyatt, 1985). Also, the studies that used general, or what Wyatt and Peters (1986) call broad funnel type questions, compared to specific or inverted funnel type questions reported lower rates (e.g., Finkelhor, 1979; Kercher & McShane, 1984; Miller, 1976). Typical inverted funnel type questions use behaviourally specific questions which may trigger memories or facilitate recollections that might not be retrieved by the more general funnel type questions. Asking specific questions allows the researcher, rather than the respondent, to decide whether the event should be classified as sexual abuse. For instance, Russell (1984) asked about specific events, such as whether the respondent was upset by anyone exposing their genitals, whether anyone tried or succeeded in any kind of sexual intercourse against the respondent's wishes, and so on. These kinds of specific prompt questions may facilitate recollection and elicit reports of sexual abuse. Thus, the type of question and the number of screening questions appear to affect prevalence rates.

Response Rates

As noted in Table 1, response rates range from 50% to 100%. It is difficult to ascertain whether response rates have any effect on prevalence rates for two reasons. First, of the 23 studies reviewed in Table 1, 9 of them did not provide response rates. Second, there is no information available as to whether those who refused to take part in the study were more likely to be victims of sexual abuse than those who participated. Thus, it is difficult to find any relationship between response rates and reported prevalence rates.

CONCLUSION

The above discussion has attempted to present a comprehensive overview of a slice of the methodological difficulties involved in trying to obtain an accurate assessment of the incidence and prevalence of the nature of child sexual abuse in the USA, Canada, Europe and Australia. Despite a variety of methodological problems present in many studies which attempt to estimate incidence and prevalence rates, most researchers believe that the rates reported in their studies are conservative figures. This is especially so since sexual abuse is such a stigmatizing phenomenon, meaning that many respondents may not like to disclose that they have been abused. Disparate definitions of child sexual abuse, and the age demarcation of a child, the nature of the historical period, and the nature of the increase of sexual abuse all seem to influence the findings. Methodological factors such as choice of sample, differences in data collection, response rates, the types and number of screening questions, and sample characteristics also appear to affect the results. The use of greater consistency in future research, particularly with respect to the definition of sexual abuse, would assist comparisons of data and allow conclusions to be drawn with less ambiguity and greater validity.

REFERENCES

Badgley, R. E, Allard, H. A., McCormick, N., Proudfoot, P., Fortin, D., Ogilvie, D., Rae-Grant, Q., Gelinas, P., Pepin, L., & Sutherland, S. (1984). Sexual offenses against children and youth (Vols. 1 & 2). Ottawa, Canada: Canadian Publishing Centre.

Bagley, C., & Ramsay, R. (1986). Sexual abuse in childhood: Psychosocial outcomes and implications for social work practice. Journal of Social Work and Human Sexuality, 4, 33-47.

Baker, A. W., & Duncan, S. P. (1985). Child sexual abuse: A study of prevalence in Great Britain. Child Abuse and Neglect, 9, 457-467.

Bell, A., & Weinberg, M. (1981). Sexual preference: Its development among men and women. Bloomington, IN: Indiana University Press.

Black, C. A., & De Blassie, R. R. (1993). Sexual abuse in male children and adolescents: Indicators, effects and treatment. Adolescence, 28, 123-133.

Bradburn, N. M. (1983). Response effects. In P. H. Rossi, J. D. Wright, & A. B. Anderson (Eds.), Handbook of survey research (pp. 289-328). New York: Academic Press.

Briere, J. (1992). Methodological issues in the study of sexual abuse effects. Journal of Consulting and Clinical Psychology, 60, 196-203.

Briem, J., & Runtz, M. (1988). Symptomatology associated with childhood sexual victimization in a nonclinical adult sample. Child Abuse and Neglect, 12, 51-59.

Briggs, F., & Hawkins, R. M. F. (1995). Protecting boys from the risk of sexual abuse. Early Child Development and Care, 110, 19-32.

Cermak, P., & Molidor, C. (1996). Male victims of child sexual abuse. Child and Adolescent Social Work, 13, 385-400.

De Francis, V. (1969). Protecting the child victim of sex crimes committed by adults. Denver, (20: American Humane Association.

De Jong, A., Hervada, A. R., & Emmeu, A. A. (1983). Epidemiologic variations in childhood sexual abuse. Child Abuse and Neglect, 7, 155-162.

Finkelhor, D. (1979). Sexually victimized children. New York: Free Press.

Finkelhor, D. (1980). Risk factors in the sexual victimization of the children. Child Abuse and Neglect, 4, 265-273.

Finkelhor, D. (1981). The sexual abuse of boys. Victimology: An International Journal, 6, 76-84.

Finkelhor, D. (1984). Child sexual abuse: New theory and research. New York: Free Press.

Finkelhor, D (1994) The international epidemology of child sexual abuse, Child Abuse and Neglect, 18, 409-417.

Finkelhor, D., Hotating, G., Lewis, I. A., & Smith, C. (1990). Sexual abuse in a national survey of adult men and women: Prevalence, characteristics, and risk factors. Child Abuse and Neglect, 14, 19-28.

Finkelhor, D., & Russell, D. E. H. (1984). Women as perpetrators: Review of the evidence. In D. Finkelhor (Ed.), Child sexual abuse: New theory and research. New York: Free Press.

Finkelhor, D., Williams, L. M., Burns, N., & Kalinowski, M. (1988). Sexual abuse in day care: A national study. Family Research Lab: University of Hampshire.

Fritz, G. S., Stall, K., & Wagner, N. N. (1981). A comparison of males and females who were sexually molested as children. Journal of Sex and Marital Therapy. 7, 54-59.

Fromuth, M. E. (1983). The long-term psychological impact of childhood sexual abuse. Unpublished doctoral dissertation, Auburn University, Auburn, Alabama.

Gagnon, J. H. (1965). Female child victims of sexual offenses. Social Problems, 13, 176-192.

Giarretto, H. (1982). Integrated treatment of child sexual abuse. Palo Alto, CA: Science & Behavior Books.

Gilgun, J. F. (1991). Resilience and the intergenerational transmission of child sexual abuse. In M. Q. Patton (Ed.), Family sexual abuse: Frontline research and evaluation (pp. 93-105). Newbury Park, CA: Sage Publication Inc.

Goldman, R. J., & Goldman, J. D. G. (1982). Children's sexual thinking. London: Routledge & Kegan Paul Ltd.

Goldman, R. J., & Goldman, J. D. G. (1988a). The prevalence and nature of the child sexual abuse in Australia. Australian Journal of Sex, Marriage and Family. 9, 94-106.

Goldman, R. J., & Goldman, J. D. G. (1988b). Show me yours: Understanding children's sexuality. Victoria, Australia: Penguin Books.

Goldman, J. D. G., & Padayachi, U. K. (1997). The prevalence and nature of child sexual abuse in Queensland, Australia. Child Abuse and Neglect, The International Journal. 21. 489-498.

Goldman, J. D. G., & Padayachi, U. K. (2000a). The perpetrators of child sexual abuse in Queensland, Australia, Children Australia, 25, XX-XX.

Goldman, J. D. G., & Padayachi, U. K. (2000b). School counsellors' knowledge of the nature of child sexual abuse and procedures and laws. Australian Journal of Guidance and Counselling XX-XX.

Gorey, K. M., & Leslie, D. R. (1997). The prevalence of child sexual abuse: Integrative review adjustment for potential response and measurement biases. Child Abuse and Neglect, 21, 391-398.

Harrold, K. (1987). Child abuse: Toward a preventive approach. Unpublished master's thesis, University of Queensland, Brisbane, Australia.

Hayman, C. R., & Lanza, C. (1971). Sexual assault on women and girls. American Journal of Obstetrics and Gynecology, 109, 480-486.

Herman, J. L. (1981a). Father-daughter incest. Cambridge, MA: Harvard University Press.

Herman, J. L. (1981b). Father-daughter incest. Professional Psychology, 12, 76-80.

James, J., Womack, W., & Stauss, F. (1978). Physicians reporting of sexual abuse of children. Journal of American Medical Association, 240, 1145-1146.

Kercher, G., & McShane, M. (1984). The prevalence of child sexual abuse victimization in an adult sample of Texas residents. Child Abuse and Neglect, 8, 495-501.

Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior in the human male. Philadelphia: W.B. Saunders.

Kinsey, A. C., Pomeroy, W. B., Martin, C. E., & Gebhard, P. (1953). Sexual behavior in the human female. Philadelphia: W.B. Saunders.

Landis, J. T. (1956). Experience of 500 children with adult sexual deviants. Psychiatric Quarterly, 30, 91-109.

Landis, C., Landis, A. T., Bolles, M. M., Metzer, H. E, Pitts, M. W., D'Esopo, D. A., Moloy, H. C., Kleegman, S. J., & Dickenson, R. L. (1940). Sex in development. New York: Paul B. Hoebert.

Leventhal, J. M. (1990). Epidemiology of child sexual abuse. In R. K. Oates (Ed.), Understanding and managing child sexual abuse (pp. 18-41). Sydney, Australia: Harcourt Brace Jovanovich.

Mazza, D., Dennerstein, L., & Ryan, V. (1996). Physical, sexual and emotional violence against women: A general practice-based prevalence study. Medical Journal of Australia, 164, 14-17.

Miller, P. (1976). Blaming the victim of child sexual molestation: An empirical analysis. Dissertation Abstract International, (University Microfilms No. UMI 77-10069).

Mrazek, P. J., Lynch, M. A., & Bentovim, A. (1983). Sexual abuse of children in the United States. Child Abuse and Neglect, 7, 147-153.

Mullen, P. E., Roman-Clarkson, S. E., & Walton, V. A. (1988). Impact of sexual and physical abuse on women's mental health. The Lancet, 1, 841-845.

National Center on Child Abuse and Neglect. (1981). Study findings: National study of incidence and severity of child abuse and neglect. Washington, DC: U.S. Department of Health and Human Services.

O'Hagan, K., (1989). Working with child sexual abuse: A post-Cleveland guide to effective principles and practice, Philadelphia: The Open University Press.

Padayachi, U. K. (1994). The nature and prevalence of child sexual abuse in Queensland and school counsellors' responses, Unpublished doctoral thesis, University of Queensland, Brisbane, Australia.

Peters, S. D., Wyatt, G. E., & Finkelhor, D. (1986). Prevalence. In D. Finkelhor (Ed.), Sourcebook on child sexual abuse (pp. 15-59). Beverly Hills, CA: Sage.

Pierce, R., & Pierce, L. H. (1985). The sexually abused child: A comparison of male and female victims. Child Abuse and Neglect, 9, 191-199.

Rind, B., & Tromovitch, P., (1997). A meta-analytic review of findings from national samples on psychological correlates of child sexual abuse. The Journal of Sex Research, 34, 237-255.

Rind, B., Tromovitch, P., & Bauserman, R. (1998). A meta-analytic examination of assumed properties of child sexual abuse using college samples. Psychological Bulletin, 124, 22-53.

Roosa, M. W., Reyes, L., Reinholtz, C., & Angelini, P. J. (1998). Measurement of women's child sexual abuse experiences: An empirical demonstration of the impact of choice of measure on estimates of incidence rates and of relationships with pathology. The Journal of Sex Research, 35, 225-240.

Russell, D. E. H. (1984). Sexual exploitation: Rape, child sexual abuse, sexual harassment. Beverly Hills, CA: Sage.

Russell, D. E. H. (1986). The secret trauma: Incest in the lives of girls and women. New York: Basic Books.

Russell, D. E. H., & Finkelhor, D. (1984). The gender gap among perpetrators of sexual abuse. In D. E. H. Russell (Ed.), Sexual exploitation: Rape, child sexual abuse, sexual harassment (pp. 215-228). Beverly Hills, CA: Sage.

Sarafino, E. P. (1979). An estimate of nationwide incidence of sexual offenses against children. Child Welfare, 58, 127-134.

Scherzer, L. N., & Lala, P. (1980). Sexual offenses committed against children. Clinical Pediatrics, 19, 683-685.

Sedney, M. A., & Brooks, B. (1984). Factors associated with a history of childhood sexual experiences in a nonclinical female population. Journal of the American Academy of Child Psychiatry, 23, 215-218.

Seidner, A. L., & Calhoun, K. S. (1984, April). Childhood sexual abuse: Factors related to differential adult adjustment. Paper presented at the second National Conference for Family Violence Researchers, Durham, NH.

Siegel, J. M., Sorenson, S. B., Golding, J. M., Burman, A. M., & Stein, J. A. (1987). The prevalence of childhood sexual assault: The Los Angeles epidemiologic catchment area project. American Journal of Epidemiology, 126, 1141-1164.

Violato, C., & Genuis, M. (1993). Problems of research in male child sexual abuse: A review. Journal of Child Sexual Abuse, 2, 33-54.

Watkins, B., & Bentovim, A. (1992). The sexual abuse of male children and adolescents: A review of current research. Journal of Child Psychology and Psychiatry, 33, 197-248.

Wyatt, G. E. (1985). The sexual abuse of Afro-American and White-American women in childhood. Child Abuse and Neglect, 9, 507-519.

Wyatt, G. E., & Peters, S. D. (1986). Issues in the definition of child sexual abuse in prevalence research. Child Sexual Abuse and Neglect, 10, 231-240.

Wynkoop, T. E, Capps, S. C., & Priest, B. J. (1995). Incidence and prevalence of child sexual abuse: A critical review of data collection procedures. Journal of Child Sexual Abuse, 4, 49-66.

Manuscript accepted February 22, 2000

Address Correspondence to Juliette D. G. Goldman, PhD, Senior Lecturer, Faculty of Education, Griffith University, PMB 50, Gold Coast Mail Centre, Queensland, Australia 4217; e-mail: [e-mail].