Title: Methodological issues in child sexual abuse research
Author(s): Jeffrey J. Haugaard and Robert E. Emery
Affiliation: Department of Psychology, University of Virginia
Citation: Haugaard, J.J. & Emery, R.E., “Methodological issues in child sexual abuse research,” Child Abuse & Neglect, vol. 13, 1989, pp. 89-100.
This study examined how changing the methods of research into child sexual abuse affects the findings of that research. College students were surveyed about their underage sexual experiences with adults. Three aspects of the study’s methods were varied: response rate, ordering of survey questions, and definition of child sexual abuse. The authors examined how these factors affected estimates of prevalence and assessments of psychological effects.
Questionnaires were distributed to 1,784 undergraduate students, and 1,089 of the students completed and returned the surveys. Their average age was 19. Sixty-one percent of the respondents were female and 39% were male. Ninety-nine percent were single, 89% were white, 6% were black, and 4% were Asian. Most were upper-middle class.
The students were asked about sexual contact with another child before age 13, sexual contact with an adult over 16 before age 13, and unwanted sexual contact after age 13. For each of these types of contact, they were asked about the type and significance of the relationship, the frequency and type of sexual activity, the amount and type of coercion, whether they told another person about the activity, their experience with social services and legal agencies, their reaction at the time, and their current reaction.
To assess current psychological functioning, the researchers administered the Comrey Personality Scales to measure students’ levels of trust and emotional stability, and the Texas Social Behavior Inventory to measure social competence. The latter examined current peer relationships and perceived social skills.
To determine effects of the ordering of questions, students were randomly divided into two groups. One group answered the questions about sexual contact before completing the personality and social functioning assessments. The other group completed these two parts in the reverse order. There were no differences in any of the results between these two groups, leading the investigators to conclude that the ordering of questions had no effect on prevalence estimates or assessed psychological effects.
To determine effects of response rates, the investigators chose three classes of students with widely varying response rates (25%, 42%, and 74%) for further study. The class with the high response rate had been motivated to participate by receiving course credit for participation. The authors administered a follow-up questionnaire with all students in these classes to determine whether they had returned the original questionnaire, and whether they had experienced childhood sexual abuse.
In the two classes with low response rates, students who had responded to the original questionnaire tended more often to be those who experienced sexual abuse, so that low response rates resulted in overestimated prevalence rates. However, in the class with the high response rate, students who had responded tended more often to be those who did not experience sexual abuse, so that prevalence rates were underestimated. The actual prevalence rates in the three classes were from 6.6% to 8.0%.
Finally, to study effects of different definitions of sexual abuse, the researchers examined three different groups of students resulting from different definitions of abuse. The first group was based on a very broad definition: any wanted or unwanted contact or non-contact sexual interaction when the student was 16 or under, and the other person was at least 5 years older and at least 16 years old.
In this group, the most common sexual activity was fondling, exhibitionism, and sexual kissing. Eighty percent of the incidents involved someone from in the family or otherwise known to the child.
The second group was obtained by removing those who rated both their response at the time and their current memory as very positive, and those who experienced only non-contact exposure by an exhibitionist.
Those experiences rated as very positive were heterosexual; one involved kissing only, five involved fondling, one involved oral sex, and three involved attempted or completed intercourse. In these cases, perceived pressure from the adult to participate in the sexual contact was significantly lower than occurred in the case of other respondents. In addition, their guilt at the time was significantly lower, and their perceived impact of the experience was significantly more positive. However, their current guilt and perceived amount of impact was similar to those of other respondents.
Those who experienced exhibitionism also felt significantly less pressured and guilty at the time of the experience than did other respondents.
The third group of students was obtained by including only those who experienced oral, anal, or vaginal intercourse. This group was intended to resemble victims of CSA seen in clinical samples.
Prevalence rates for each definition of CSA are shown below.
|Broad definition||Narrow definition||Clinical type|
The researchers assessed the level of psychological functioning (trust, emotional stability, peer relationships, and social skills) of each group, and compared it with that of a control group of 85 students who had no childhood sexual experience with adults. There were no clear-cut distinctions between the broadly and narrowly defined groups, but all differences were in the direction of the narrowly defined group showing more negative differences from the control group than the broadly defined group showed.
The broadly defined group exhibited no statistically significant differences from the control group. The narrowly defined group showed marginally significant differences from the control group in just two areas: poorer same sex relationships and less emotional stability. The “clinical” group showed statistically significant differences from the control groups in two areas—poorer peer and sexual relationships—and marginally significant differences in three areas: lower levels of trust, less emotional stability, and poorer perceived social skills.
In the broadly defined group, there were significant gender differences: males felt less guilty and more positive about the experience at the time, had more positive memories of the experience, and felt less pressure to participate. However, in the narrowly defined group, there were no significant gender differences.
The authors conclude that the definition of sexual abuse significantly influences prevalence estimates and has an effect on findings related to consequences of CSA. Narrower definitions are associated with more negative consequences. Broad definitions may have the effect of erroneously “minimizing experiences of some victims while raising overconcern about the experiences of others.”
The authors argue for a “commitment to providing reasoned, clear definitions of child sexual abuse.” Which definition is selected depends on the writer’s audience and purpose. However, the dangers of tailoring the definition to the audience include a muddled concept of child sexual abuse, difficulty in comparing studies, and misled readers who may assume a particular definition is being used when it is not.