The national samples were useful in examining popular assumptions about CSA. Some of their shortcomings, however, were that there are very few of these studies, these studies have very little data on reactions, and inadequate information to judge the assumption of causality of harm. We thus conducted a second meta-analysis based on another group of nonclinical samples--college samples. We chose college samples for several reasons. One is that these represent the largest number of nonclinical samples of the same kind. Despite the fact that persons with a college background are different from those without, we felt college samples would be useful toward answering questions about population characteristics--that is, how does the typical person with CSA react to it--because, in the U.S., at least 50% of the adult population has had some college exposure.
Another value of college samples is that these studies have generally been conducted by university researchers, who have designed their studies well, often taking into account family environment factors. This information, not systematically available in clinical studies or even the national studies, is useful for examining the causal role that CSA might play in producing negative effects. Additionally, these studies have provided a rich source of data for examining reactions to CSA experiences, not well provided in the other literature. This information is useful for examining assumptions about CSA such as pervasiveness and intensity of effects, as well as gender equivalence in reactions.
Altogether, we obtained 59 usable studies for examining CSA-adjustment relations, reactions, and self-reported effects. In examining the relation between CSA and adjustment, 54 samples were used, which included 3,254 male subjects from 18 samples and 12,570 female subjects from 40 samples. Reactions and self-reported effects were based on 783 male subjects from 13 samples and 2,353 female subjects from 14 samples.
Definitions of CSA varied across these studies. For example, 20% restricted their definition to include only unwanted CSA experiences. The remaining 80% included both willing and unwanted CSA experiences, and most often defined CSA as an age difference between partners of 5 or more years where the younger partner was less than 16 or 17 years of age. Prevalence rates for CSA, based on the various definitions, were as follows. For males, based on 26 samples with 13,704 subjects, CSA was reported 14% of the time. For females, based on 45 samples with 21,999 subjects, CSA was reported 27% of the time.
Some researchers have argued that data from college samples are not informative about the effects of the more severe forms of CSA, because college subjects experience less severe forms of CSA than do people in the general population. By going back to the national samples and pulling out the relevant data, and by going through the college samples and computing corresponding values, we were able to test this assumption.
Table 4 shows some of these results. It has been argued that severity increases from noncontact CSA, such as exhibitionism, to fondling, to oral sex, to intercourse. In the table, you can see that college subjects had just as much intercourse as national subjects--and much more in the case of males. Relatedness between the younger and older participants has also often been used as an indicator of severity, with incestuous contacts seen as the most severe.
Table 5 shows that college subjects experienced just as much incest as persons in the general population.
Another commonly used indicator of severity is frequency of CSA occurrences--that is, multiple episodes are viewed as more severe than single episodes. In both the college and national samples, about half of those who had CSA had multiple episodes, showing once again similarities in terms of severity. Our conclusion from these comparisons is that, because CSA characteristics are nearly the same in both college and national samples, using college samples to answer questions about CSA in the general population seems well justified.
Prevalence Rate Estimates of Four Types of CSA in College and National Populations
. k is the number of samples and N is the number of SA respondents in these samples that prevalence rate estimates of types of CSA are based on. Prevalence rate estimates are weighted means of prevalences from individual samples. College estimates come from studies included in the current review; national estimates come from 3 studies of national samples (Baker & Duncan, 1985; Laumann et al., 1994; Lopéz et al., 1995)
In some college and national studies, intercourse included both attempted and completed acts
(b) Combined values were based on two additional studies (with a male and female sample in each) that reported only combined results
(c) For exhibitionism, only data from Lopéz et al. were reported (female: k=1, N=203; male k=1, N=134; combined k=2, N=337); for oral sex, only data from Laumann et al. and Lopéz et al. were reported (female: k=2, N=476; male: k=2, N=291; combined k=4, N=767).
Prevalence Rate Estimates of Relationship Between CSA Respondents
. Close family CSA includes sexual relations with very close relatives (e.g., biological or step parents, grandparents, older siblings). Wider family CSA includes both close family CSA and relations with other relatives. Prevalence rate estimates are weighted means of prevalences from individual samples. College estimates come from studies included in the current revies; national estimates come from 3 studies of national samples (Baker & Duncan, 1985; Laumann et al., 1994; Lopéz et al., 1995)
Based on 21, 9, and 33 samples for females, males, and combined, respectively
b Based on 3, 3, and 6 samples for females, males, and combined, respectively
c Based on 10, 6, and 19 samples for females, males, and combiend, respectively.
We next examined the relationship between CSA and adjustment by meta-analyzing results across the 54 samples that provided usable statistics. Based on 15,912 subjects, the average amount of variability in adjustment scores accounted for by CSA was 0.81%, meaning that CSA failed to account for 99.19% of the adjustment variability. Nevertheless, this small difference in adjustment was statistically significant, with CSA subjects showing somewhat poorer adjustment. We next meta-analyzed the relations between CSA and adjustment separately for males and females. As you can see in Table 6, CSA accounted for 0.49% of the adjustment variability for males and 1.00% for females--exactly the same values as in the national samples. It is worth emphasizing at this point that the comparability of the college samples and the national samples is quite good in various respects: prevalence rates of CSA, types of CSA, and the magnitude of the CSA-adjustment relations. These findings indicate that college data are substantially more valuable than clinical data for attempting to understand the nature of CSA in the general population.
Meta-Analyses of CSA-Adjustment Relations in College Students for Males and Females
Note. k represents the number of samples; N is the total number of subjects in the k samples;
% variance stands for the percent of variability in adjustment that CSA accounts for.
Because a sizable minority of the studies restricted their definitions of CSA to unwanted sex only, we took the opportunity to examine relations between CSA and adjustment as a function of level of participation. We did this separately for males and females. Table 7 shows the results. For males, when just considering samples that included all types of CSA (that is, both willing and unwanted sex), we found that CSA accounted for only 0.16% of the adjustment variability, which was not statistically significant. When just examining samples where the CSA was unwanted, CSA accounted for 1.69% of the adjustment variability, which was statistically significant. This value was greater than the previous value for both willing and unwanted sex by a factor of 10.
Taken together, these two results imply that, for boys, willingly engaging in CSA is not associated with poorer adjustment. For females, on the other hand, CSA was associated with poorer adjustment whether both willing and unwanted CSA were considered together or unwanted CSA only was considered. In the former case, CSA accounted for 1.21% of the adjustment variability; in the latter, it accounted for 0.64%. We compared the four effect sizes for these four conditions and found that the effect size for males in the willing and unwanted combined condition was statistically significantly smaller than the effect sizes in the other three conditions, which were all statistically equivalent. This finding points to a sex difference, and implies that willing boys should not be grouped with girls when discussing the effects of CSA.
Meta-Analyses of CSA- Adjustment Relations in College Students for Each Gender by Consent Combination
. k represents the number of samples; N is the total number of subjects in the k samples, % variance stands for the percent of variability in adjustment that CSA accounts for.
All types of consent included both willing and unwanted CSA; unwanted CSA includes unwanted experiences only.
So, at least for boys, we see that CSA has no inevitable outcome, but depends on the context in which it occurs. To examine context further, we focused just on subjects in the college samples who had CSA to see what factors might or might not be related to their reactions or symptoms. The contextual factors we examined were the frequency of CSA episodes, their duration over time, the use of force, whether penetration occurred, and whether the CSA was incestuous.
Table 8 shows the results of our analyses. Contrary to popular assumptions, reactions were not more negative, and symptoms were not greater, with greater frequency of episodes, longer duration of these relationships, or the presence of penetration. On the other hand, the use of force and incestuous relations were related to more negative reactions and more symptoms.
Meta-Abalyses of Relations Between Aspects of the CSA Experience and Outcome In CSA College Students
Note. k represents the number of samples; N is the total number of subjects in the k samples with CSA experiences; % variance stands for the percent of variability in reactions/effects or symptoms that the moderator accounts for among the CSA subjects. Values in parentheses indicate the moderate was related to less negative reactions/effects or symptoms.
* indicates a statistically significant result
The image of CSA as portrayed in the media is that of a frail, helpless child in a state of shock after having been ravaged by an adult. We next present data relevant to assessing the validity of this image. Table 9 presents results from 10 female samples and 11 male samples on how subjects reacted, at the time, to their CSA experience. Of the 1,421 female experiences of CSA, 11% were positive, 18% were neutral, and 72% were negative. Of the 606 male experiences, 37% were positive, 29% were neutral, and 33% were negative. The results for males strongly contradict the popular image just described. The majority of boys (two-thirds) did not react negatively. For girls, the pattern was just the reverse, showing a striking sex difference. This once again provides evidence against the assumption of gender equivalence--that boys and girls react the same. In terms of negative reactions, it is important to note that such reactions can range from mild discomfort to traumatic shock. The percentages of boys and girls who react in accord with the popular image of traumatic shock would be only a fraction of the figures just presented for negative reactions.
Retrospectively Recalled Immediate Reactions of College Students to their CSA Experiences
Note. n/a indicates information not available. Totals include only samples for which all 3 reaction-types are given. Total percents are weighted by sample size; total Ns reflect a combination of number of experiences and number of subjects. Percents do not sum exactly to 100 because of rounding.
a Includes mixed reactions.
b Indicates number of experiences. Otherwise, N indicates number of subjects.
We dont know what fraction this is, but presumably traumatic shock would result in self-perceived negative effects, probably of a lasting nature. We examined self-perceived effects across the college samples to address this issue. Table 10 shows the results for the studies that contained this information. Self-perceived lasting negative effects were uncommon for males.
In Condys study, only 16% of male subjects with CSA felt that this experience had negatively affected their current sex lives. In Fishmans study the corresponding value was 13%, in Fritzs study it was 10%, in Landis study it was 0.4%, and in West and Woodhouses study only one or two out of 67 felt a current negative impact on their sex lives. In terms of other types of lasting effects, in Landis study, none of the males felt there was any permanent harm to their emotional development. In Fishmans study, about a quarter of the male subjects felt some negative impact on their overall life.
Self-Reported Effects of CSA Experiences on College Students
a Indicates number of experiences. Otherwise, N indicates number of subjects.
Females differed from males in reported higher rates of perceived negative effects. In terms of lasting effects on their sex lives, the figures were 24% in Fritzs study and 2.2% in Landis study. In terms of lasting negative effects of a more general nature, in Hrabowys study 20% felt moderately troubled over it now and 5% felt very troubled. In Landis study, 4% felt they still had not recovered. In Nash and Wests study, 22% felt they were still negatively affected. These data, both for males and females, show that some subjects were seriously harmed by their experience, but only a minority. These data suggest that CSA has the potential to be harmful, but that serious harm is not an innate property of CSA. As our analyses presented previously suggest, it is the context, such as the use of force, or the lack of willingness on the younger persons part, or potentially incestuous involvement, that combines with CSA to produce harmful outcomes.