Beruflich Dokumente
Kultur Dokumente
The Puzzle of Intrafamilial Child Sexual Abuse: A Meta-Analysis Comparing
Intrafamilial and Extrafamilial Offenders with Child Victims
PII: S0272-7358(15)00056-2
DOI: doi: 10.1016/j.cpr.2015.04.001
Reference: CPR 1433
Please cite this article as: Seto, M.C., Babchishin, K.M., Pullman, L.E. & McPhail, I.V.,
The Puzzle of Intrafamilial Child Sexual Abuse: A Meta-Analysis Comparing Intrafa-
milial and Extrafamilial Offenders with Child Victims, Clinical Psychology Review (2015),
doi: 10.1016/j.cpr.2015.04.001
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 1
The Puzzle of Intrafamilial Child Sexual Abuse: A Meta-Analysis Comparing Intrafamilial and
Michael C. Seto
PT
RI
Kelly M. Babchishin
SC
University of Ottawa Institute of Mental Health Research
NU
Lesleigh E. Pullman
MA
University of Ottawa
D
TE
Ian V. McPhail
P
CE
University of Saskatchewan
AC
Authors‘ Note
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 2
Please address correspondence to Michael Seto, Director, Forensic Research Unit, University of
Ottawa Institute of Mental Health, 1804 County Road 2 East, Brockville, ON, Canada, K6V
PT
5V8. Telephone: (613) 345-1461 ext 2605; Fax: (613) 345-7276. Email:
RI
michael.seto@theroyal.ca
SC
Thank you to Marcus Boccaccini, R. Karl Hanson, Yolanda Fernandez, Genevieve Martin, Jill
Levenson, and Wineke Smid who took the time to share unpublished data with us and/or
NU
answered our questions. We are also grateful to Karl Hanson and Martin Lalumière for their
MA
helpful comments on an earlier version of this paper.
D
P TE
CE
AC
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 3
Abstract
Intrafamilial child sexual abuse is a serious social and health problem, yet explanations of sexual
offending against children that emphasize antisocial tendencies and atypical sexual interests do
PT
not adequately explain intrafamilial offending. In this meta-analysis, we tested other
RI
explanations of intrafamilial child sexual abuse by examining 78 independent samples that
SC
studies disseminated between 1978 and 2013 (Mdn = 2000). Intrafamilial offenders were
NU
significantly lower on variables reflecting antisocial tendencies (e.g., criminal history, juvenile
delinquency, impulsivity, substance use, and psychopathy) and atypical sexual interests (e.g.,
MA
pedophilia, other paraphilias, excessive sexual preoccupation). Contrary to other explanations
that have been proposed, intrafamilial offenders scored lower on offense-supportive attitudes and
D
beliefs, emotional congruence with children, and interpersonal deficits; intrafamilial offenders
TE
also did not differ from extrafamilial offenders on most indicators of psychopathology.
P
Intrafamilial offenders were, however, more likely to have experienced sexual abuse, family
CE
abuse or neglect, and poor parent-child attachments. There were too few studies to examine
AC
family dynamics – spousal relationship quality, parent-child victim relationship, and family
functioning more generally – even though these factors have been frequently mentioned in the
clinical and theoretical literatures. Implications for theories of intrafamilial sexual offending,
Keywords: incest; intrafamilial sex offenders; sexual offending against children; meta-analysis
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 4
Introduction
PT
Intrafamilial child sexual abuse is a widespread social and health problem. Stoltenborgh,
RI
van Ijzendoorn, Euser, and Bakersmans-Kranenburg (2011) conducted an extensive meta-
analysis on the prevalence of child sexual abuse across the world, updating meta-analyses by
SC
Pereda, Guilera, Forns, & Gómez -Benito (2009) and Finkelhor (1994). Based on 331
NU
independent samples and almost 10 million individuals, Stoltenborgh et al. found an overall
prevalence rate of 13%, with the rate for girls being more than twice that of boys (18% vs. 8%,
MA
respectively). Up to one-third of child sexual abuse cases are perpetrated by family members. Of
these intrafamilial cases, fathers and step-fathers are the most common type of relative, typically
D
TE
representing the majority of convicted intrafamilial offenders (Gibbens, Soothill, & Way, 1978;
Rice & Harris, 2002; Richards, 2011; Seto, Lalumière, & Kuban, 1999).
P
Child sexual abuse is associated with a range of mental and physical health costs, though
CE
fortunately not for all victims (Beitchman et al., 1992; Rind, Tromovitch, & Bauserman, 1998);
AC
victims of child sexual abuse committed by non-relatives (Stroebel et al., 2012). Longitudinal
studies have found that victims of child sexual abuse have poorer psychological well-being,
greater rates of teenage pregnancy, higher likelihood of multiple victimization, and lower
socioeconomic status in adulthood than non-victims (Robert, O‘Connor, Dunn, & Golding, 2004;
Widom, Czaja, & Dutton, 2008). Sexual abuse does not only affect the lives of the victims;
children of abuse survivors have greater adjustment problems than children of women who have
not been sexually victimized in childhood (Robert et al., 2004). In addition to the personal costs,
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 5
costs to society include health costs when survivors of childhood sexual abuse require health care
as adults (e.g., Chen et al., 2010; Paras et al., 2009), with adult women who experienced sexual
abuse as children having annual health costs that are 16% higher than women without a sexual
PT
abuse history (Bonomi et al., 2008).
RI
Explaining Sexual Offending Against Children
Attempts to prevent intrafamilial child sexual abuse will be most effective when based on
SC
a sound understanding of the motivations behind these offenses. We can begin by looking at the
NU
broader literature on child sexual abuse. Key candidates for the initiation and maintenance of
sexual offending against children fall broadly into two dimensions labelled atypical sexual
MA
interests and antisocial tendencies (Seto, 2008, 2013). Atypical sexual interests include
sexual attraction to pubescent children), as well as excessive sexual preoccupation. These aspects
of sexuality can be motivations for sexual behavior involving children. Antisocial tendencies
P
include impulsivity, callousness, risk-taking, antisocial attitudes and beliefs, and a pattern of
CE
unstable, irresponsible behavior. These aspects of cognition and personality represent facilitators
AC
of sexual offending against children, because an antisocial individual is more willing to act on
his sexual motivations and is less sensitive to the risk of that action, including negative
what is known about factors that predict sexual recidivism (Hanson & Morton-Bourgon, 2005;
Seto, 2008, 2013). Individuals who are high on antisocial tendencies are at higher risk of sexual
or nonsexual recidivism, while individuals who are high in atypical sexual interests are at higher
risk specifically of sexual recidivism. Those who are high on both risk dimensions are at the
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 6
greatest risk of sexually reoffending (Hawes, Boccaccini, & Murrie, 2013; Seto, 2008).
PT
intrafamilial offenders typically score lower on measures of antisocial tendencies (e.g., Rice &
RI
Harris, 2002; Seto & Barbaree, 1999) and show less or comparable sexual arousal to children
compared to extrafamilial offenders (Abel, Becker, Murphy, & Flanagan, 1981; Chaplin, Rice, &
SC
Harris, 1995; Frenzel & Lang, 1989; Freund, Watson & Dickey, 1991; Lang, Black, Frenzel, &
NU
Checkley, 1988; Quinsey, Chaplin, & Carrigan, 1979). In other words, neither antisocial
tendencies nor atypical sexual interests are common among intrafamilial offenders, even though
MA
intrafamilial offenders tend to have younger victims and offend for longer periods of time
compared to extrafamilial offenders (Seto, 2008). Of course, some intrafamilial offenders are
D
TE
highly antisocial, and some show evidence of pedophilia; however, the dimensions of
antisociality and atypical sexual interests do not explain why some men sexually offend against
P
Intrafamilial child sexual abuse is also deeply puzzling from a biological and sociological
AC
perspective, given evidence of the negative effects of close inbreeding on offspring, incest
avoidance, and a strong incest taboo across cultures and times (see Seto, 2008; Thornhill, 1990;
Westermarck, 1891/1921; Wolf & Durham, 2004, for review). Briefly, incest avoidance refers to
the avoidance of sexual behavior among close relatives (e.g., siblings) observed among humans
and other animals (e.g., Koenig, Haydock, & Stanback, 1998), and incest taboo refers to the
cultural rules or norms that discourages sexual relationships between relatives. Though the rules
regarding which familial relationships are prohibited vary, most cultures have explicit rules
about sexual behavior between genetic relatives (Thornhill, 1991). The specific mechanisms and
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 7
the extent to which incest avoidance is driven by innate versus cultural processes have been
disputed, but it is well accepted that incest avoidance exists (see Leavitt, 2005, for review). The
idea of sex with a close relative elicits disgust in most people (incest aversion: e.g., Antfolk,
PT
Karlsson, Bäckström, & Santtila, 2012), is prohibited by religious and secular laws (though the
RI
types of prohibited relationships vary), and occurs in only a minority of families, despite ample
opportunities given close proximity and time (Finkelhor, 1994; Stoltenborgh et al., 2011).
SC
Psychological Theories of Intrafamilial Child Sexual Abuse
NU
For the many intrafamilial offenders who are low in antisocial tendencies and atypical
sexual interests, what other factors explain intrafamilial child sexual abuse? Despite the
MA
occurrence of intrafamilial child sexual abuse across cultures and time, and despite the social and
individual costs associated with this form of sexual offending, there are surprisingly few
D
TE
children, with the implicit suggestion that, for intrafamilial offenders, these risk factors are
CE
example, Seto‘s (2008, 2013) motivation-facilitation model suggests that individuals who are
sufficiently high in motivation to have sexual contact with children (such as having pedophilia
combined with a high sex drive) – and who are sufficiently high in antisocial tendencies to
facilitate acting on that motivation despite any personal and social prohibitions against the
behavior – would also be expected to be more likely to offend against a related child. Yet as we
have just discussed, this is not the case for most intrafamilial offenders, who are relatively low
on antisocial tendencies and atypical sexual interests compared to other sex offenders.
emotional congruence with children, interpersonal deficits, and disinhibition as key motivators of
sexual offending against children. In this model, someone is more likely to offend against a
related child if they score high on one or more of these risk factors and opportunities to offend
PT
arise (e.g., the person is left alone with a child for an evening).
RI
Clinically-derived explanations of intrafamilial child sexual abuse, which focus on family
structure and family member dynamics as key contributing factors, are notably different (e.g.,
SC
Maddock & Larson, 1995). Hypotheses about incestuous families have implicated the following:
NU
offending fathers or step-fathers are likely to take on an authoritarian, patriarchal role; the
marital relationship is aloof and characterized as low in sexual intimacy and/or high conflict;
MA
mothers are more likely to be dependent on the father, financially or otherwise; some daughters
are pushed into a surrogate partner role, not only sexually but also with regard to intimacy and
D
TE
household tasks, such as supervision and care of younger children (see Seto, 2008, for review).
offenders, with easier access to children. Despite greater access to potential victims, however,
AC
incest occurs relatively infrequently in families (Stoltenborgh et al., 2011). As such, access to
children is not a sufficient condition for intrafamilial sexual abuse to occur. Motivational and
facilitatory factors are necessary as well, though perhaps at lower levels than for extrafamilial
offenders who may have less access and offending opportunity. Narrative reviews have
concluded that intrafamilial offenders typically score lower on measures of atypical sexual
interests such as pedophilia and lower on measures of antisocial tendencies than extrafamilial
offenders (e.g., Seto, 2008). Given that many intrafamilial offenders are less antisocial and
sexually atypical compared to sex offenders with unrelated child victims, what other factors can
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 9
attitudes and beliefs, emotional congruence with children, childhood difficulties with sexual
PT
abuse, poor attachment to parents, interpersonal deficits, and psychopathology are important
RI
factors in explaining intrafamilial child sexual abuse. The common theme across the first four
explanations is that intrafamilial offenders turn to children to meet their sexual or emotional
SC
needs, possibly because for some, they themselves were victims of incest. The common theme
NU
across the next two explanations is that intrafamilial offenders are less able to pursue sexual
opportunities outside the family due to interpersonal deficits or psychopathology and, hence,
MA
―keep it in the family,‖ because they do not have the opportunity to offend against unrelated
victims. These potential explanations – many of which are changeable and thus potential targets
D
identify factors such as beliefs that are supportive of child sexual abuse (e.g., that children are
CE
not harmed or can even benefit from sex with an adult) and sexual entitlement (e.g., ―a person
AC
should have sex when it is needed‖) as risk factors (Finkelhor, 1984; Hanson, Harris, Scott, &
Helmus, 2007). Several studies suggest intrafamilial offenders do score higher in sexual
entitlement and offense-supportive attitudes and beliefs than extrafamilial offenders (Hartley,
1998; Pemberton & Wakeling, 2009; Wakeling, Webster, Moulden, & Marshall, 2007), as well
as nonoffending controls and male batterers (Hanson, Gizzarelli, & Scott, 1994). These kinds of
offense-supportive attitudes and beliefs are a subset of more generally antisocial attitudes and
beliefs about crime, law, and the rights of others. As such, these are psychological constructs
where the general pattern (i.e., intrafamilial offenders scoring lower on antisocial tendencies than
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 10
Emotional congruence with children. Some sex offenders are more comfortable with
children than adults. A recent meta-analysis found that intrafamilial offenders generally score
PT
lower than extrafamilial offenders on measures of emotional congruence with children (McPhail,
RI
Hermann, & Nunes, 2013). McPhail and colleagues (2013) suggest that the lower level of
emotional congruence with children in intrafamilial offenders may reflect the intrafamilial
SC
offenders‘ physical or emotional absence during the first formative years. Consistent with this
NU
interpretation, Fisher, Beech, and Brown (1999) found that intrafamilial offenders scored
Cycle of sexual abuse. Two recent meta-analyses found that adolescent and adult sex
D
TE
offenders were substantially more likely to have sexual abuse histories than other adolescent or
adult offenders (Jespersen, Lalumière, & Seto, 2009; Seto & Lalumière, 2010). Some studies
P
have suggested that intrafamilial offenders are likely to have been sexually abused, and that
CE
intrafamilial sexual abuse can ―run in families‖ (e.g., Faller, 1989). Indeed, some families exhibit
AC
perpetrated by their parents or other adult relatives grow up and then commit intrafamilial sexual
suggests that familial factors (whether cultural or genetically transmitted) can influence the
likelihood of incestuous behavior. Williams and Finkelhor (1990), for example, conducted a
review of studies published mostly in the 1980s and concluded incestuous fathers were less
involved in child care, more socially isolated, less assertive, and less interpersonally sensitive
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 11
than non-abusive fathers or step-fathers. Lu and Lung (2012) found Taiwanese intrafamilial
offenders scored lower on perceived parental care than their non-intrafamilial offending
counterparts. In addition, McKillop, Smallbone, Wortley, and Andjic (2012) found that offenders
PT
whose first sexual contact with a child was with a family member had a tendency to have more
RI
insecure paternal (65% vs. 58%) and maternal (55% vs. 46%) attachment than offenders whose
SC
Interpersonal deficits. Sexual offending against children has also been attributed to
NU
interpersonal deficits that block age-appropriate intimate relationships (Finkelhor, 1984; Hall &
Hirschman, 1992; Ward, Polaschek, & Beech, 2006). There is some evidence to support that
MA
these factors are applicable to intrafamilial offenders. For example, Bogaerts, Vanheule, and
Declercq (2005) found that intrafamilial offenders had greater levels of interpersonal problems
D
TE
than extrafamilial offenders. However, McKillop et al. (2012) did not find that intrafamilial and
extrafamilial offenders differed in adult attachment, which was typically insecure in both groups
P
(73% and 80%, respectively). Underhill, Wakeling, Mann, and Webster (2008) also found that
CE
intrafamilial offenders did not differ from other sex offenders against children on openness to
AC
women or openness to men; however, sex offenders against children as a group were lower on
offending against related children. Firestone, Dixon, Nunes, and Bradford (2005), for example,
found that intrafamilial offenders who victimized younger children were more psychiatrically
disturbed, as assessed using the Brief Psychiatric Rating Scale, and more likely to engage in
substance abuse than those who victimized older related victims. In a sample of sexual offenders
with child victims, including intrafamilial offenders, Kalichman (1991) found that victim age
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 12
was inversely correlated with psychopathology, such that those who offended against
prepubescent children were more psychiatrically disturbed on self-report measures than those
PT
The Current Meta-Analysis
RI
Past studies of intrafamilial offenders often do not include suitable comparison groups
(see Garber & Hollon, 1991). It is not enough to show, for example, that intrafamilial offenders
SC
are likely to have been sexually abused. A more informative question is whether intrafamilial
NU
offenders are more likely to have such histories compared to extrafamilial sexual offenders
without related victims. Group differences suggest the factor may be a specific causal candidate
MA
for intrafamilial sexual offending, to be pursued in longitudinal and quasi-experimental studies.
this question: Among individuals who commit a sexual offense involving a child, what
distinguishes those who offend against a related child from those who offend against an
P
unrelated child? The current meta-analysis included a wide range of variables identified as
CE
important for either sexual offending against children more generally, or intrafamilial sexual
AC
abuse specifically.
extrafamilial offenders against children, and thereby highlight potential explanations for why
some men offend against related children rather than unrelated children. The current meta-
analysis also identified domains that may be specific to intrafamilial sexual offending and, as
such, might need to be considered in the assessment and treatment of intrafamilial offenders. We
first compared intrafamilial and extrafamilial offenders to estimate the magnitude of observed
differences in indicators of antisocial tendencies and atypical sexual interests. We then compared
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 13
the two groups on other theoretically relevant domains, specific hypotheses include:
PT
tendencies and atypical sexual interests.
RI
2. If explanations that emphasize patriarchal attitudes and beliefs or sexual entitlement are
correct, intrafamilial offenders should score higher on measures of these variables than
SC
extrafamilial offenders.
NU
3. Given the results of a recent, comprehensive meta-analysis by McPhail et al. (2013), we
expected to replicate their finding that intrafamilial offenders should score lower on
MA
emotional congruence with children than extrafamilial offenders. This hypothesis is
child attachment are correct, then intrafamilial offenders should score higher on these
P
offenders.
Method
Selection of Studies
Online searches for studies comparing intrafamilial to extrafamilial sexual offenders were
conducted through PsycINFO, ProQuest Dissertations and Theses, National Criminal Justice
Reference Service, Web of Science, and Medline using the following search terms: keywords
(child* molest* or sex* offen* or paedo* or pedo*) AND (incest*, intrafam*, or extrafam*).
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 14
Additional studies were also found by reviewing the reference lists of collected studies,
conference proceedings from the Association for the Treatment of Sexual Abusers, contacting
active researchers, and utilizing Google Scholar. The search ended March 31st, 2013 and
PT
resulted in 118 eligible studies, representing 78 unique, non-overlapping samples (3 French and
RI
75 English samples). Table 1 presents the descriptive information of the included samples.
The sample size ranged from 11 to 1,052 for intrafamilial offenders (M = 84.7, total N =
SC
6,605) and from 10 to 2,805 for extrafamilial offenders (M = 135.6, total N = 10,573). The
NU
studies were published from 1978 to 2013 (Mdn = 2000). Of these 78 samples, 42 (54%) were
classified as published, while 36 (46%) were classified as unpublished. The samples were
MA
primarily drawn from the United States (k = 32) and Canada (k = 26). Most samples (95%)
grouped offenders into their respective groups based on official charges or convictions; however,
D
TE
some samples also used self-report or other sources of information such as accusations (27%).
Many samples did not report whether the intrafamilial offenders also had unrelated victims
P
(40%, k = 31); when this information was reported, most samples reported exclusively related
CE
victims (87%, k = 41), whereas the rest reported that some intrafamilial offenders also had
AC
extrafamilial victims (13%, k = 6). The majority of studies (68%, k = 53) did not indicate if the
intrafamilial offenders include sociolegal intrafamilial offenders (those who are sociolegally
related to their victim, within a nuclear family, such as stepfathers). When available, only two
samples were purely biological intrafamilial offenders (3%); most samples of intrafamilial
offenders were a mix of sociological and biological intrafamilial offenders (30%, k = 23). Studies
also tended not to report whether the extrafamilial offenders sampled also had related victims
(49%, k = 37). When this information was reported, most samples included extrafamilial
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 15
offenders who also had related victims (63%, k = 26) and 37% only sampled extrafamilial
Twenty of the 78 samples (26%) reported the average number of victims for each
PT
offender group. Extrafamilial offenders had a greater average number of victims (ranging from
RI
1.6 to 26.3; Mdn = 3.6) than intrafamilial offenders (ranging from 1.1 to 9.6, Mdn = 1.7; a
median d of 0.56). The largest average number of victims for both groups was found in Gould
SC
(1994), and was based on the sum of the average number of victims in prior and in the index
NU
offenses. Thirty-one of the 78 samples also presented information on the gender of the victims,
which was either based the index offense (k = 12) or the complete victim pool (k = 8); eleven
MA
studies did not specify how gender of the victims was coded. Intrafamilial offenders were more
likely to have girl victims than extrafamilial offenders (Mdn = 87% compared to 63%, k = 31).
D
TE
The age categories of victims were not consistently reported across samples. When this
information was available (40 of the 78 samples), samples reported mostly prepubescent victims
P
(i.e., under the age 12; k = 32), followed by a mix of prepubescent and pubescent victims (i.e.,
CE
most victims under the age of 15; k = 6), and, finally, only a few samples (k = 2) reported mostly
AC
Most samples were derived from institutions (61%; k = 45) and the majority of studies
exclusively sampled men (96%, k = 75); three samples also included a very small number (1%)
of women offenders. Of the samples in which treatment status was known (k = 59), half of
samples of intrafamilial offenders did not receive treatment (54%, k =32), 29% (k = 17) received
treatment, and 17% (k = 10) were mixed in their participation in treatment. Similarly, half of the
samples of extrafamilial offenders did not receive treatment (56%, k = 33), 27% (k = 16)
received treatment, and 17% (k = 10) were mixed in their participation in treatment. Of the
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 16
samples in which the degree of adversarial setting could be estimated (k = 71), samples were
predominantly (46%) drawn from moderately adversarial settings (e.g., where results might
influence treatment decision-making.) Twenty-one samples (30%) were drawn from low
PT
adversarial settings (e.g., voluntary clients with no foreseeable legal consequences) and 17
RI
samples (24%) were drawn from high adversarial settings (e.g., assessments used explicitly for
SC
Coding Procedure
NU
To be included in the current meta-analysis, the study had to include a comparison
between intrafamilial and extrafamilial offenders, report on at least one of the characteristics
MA
targeted by this review – which included demographic, childhood history, antisocial tendencies,
atypical sexual interest, and other psychological variables – and include sufficient statistical
D
information to calculate an effect size (Cohen‘s d). Assignment of variables to domains was
TE
informed by previous meta-analyses of the sexual offender literature, such as studies examining
P
the role of antisocial tendencies and atypical sexual interests in the prediction of sexual
CE
recidivism (Hanson & Bussière, 1998; Hanson & Morton-Bourgon, 2005). Explanations of
AC
Each study was coded with a standard list of variables and explicit coding rules; a coding
manual with standard variable list and coding rules is available upon request. The second author
coded all studies to ensure consistency across studies. Half of the studies were randomly
assigned to the third and fourth authors for interrater analyses and to generate consensus ratings,
with the exception of the 3 French samples, which were coded only by the second author.
Ratings had two components: Information describing the study (one form per study) and effect
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 17
size information (one form per effect size). At the end of coding, only variables with three or
PT
excluded the 3 French samples, 11 practice cases, and 2 samples which were coded after the
RI
interrater analyses. The raters coded 954 common effect sizes, with high levels of agreement
(absolute intra-class correlation [ICC] based on one-way random and single measure = .88).
SC
Interrater reliability ranged from 66% to 100% agreement for categorical variables (Mdn = 87%,
NU
n = 21; κ ranged from .44 to 1.00, Mdn = .73, n = 21) and ICC values ranged from .50 to 1.00 for
Effect size. The effect size indicator was the standardized mean difference (Cohen‘s d)
D
TE
and is defined as follows: d = (M1 - M2)/Sw, where M1 and M2 are the group means, and Sw is the
Sw=
N1 1 N 2 1
AC
According to J. Cohen (1988), d values of 0.20 are considered ―small,‖ 0.50 ―medium,‖
and 0.80 ―large.‖ In most cases, d was calculated using means and standard deviations. When d
was calculated from 2 by 2 tables, however, the variance of d was estimated using formula 19
from Sánchez-Meca, Marín-Martínez, and Chacón-Moscoso (2003), with 0.5 added to each cell
to permit the analysis of tables with empty cells (Fleiss, 1994). A positive d indicates that
extrafamilial sexual offenders had more characteristics that were problematic (e.g.,
Aggregation of findings. Findings across studies were aggregated using both fixed-
effect and random-effects meta-analysis (Borenstein, Hedges, Higgins, & Rothstein, 2009).
Whereas the results of fixed-effect meta-analysis are conceptually restricted to the particular set
PT
of studies included in the meta-analysis, random-effects meta-analysis estimates effects for the
RI
population of which the current sample of studies is a part. When variability across studies is low
SC
results. When the analysis includes a small number of studies (k < 30), greater interpretive
NU
weight should be given to fixed-effect rather than random-effects analyses because the between-
study variability estimate necessary for random-effects analyses loses precision (Schulze, 2007).
MA
To test the variability of findings across studies, we used Cochran‘s Q statistic and the I2
statistic (Borenstein et al., 2009). The Q statistic provides a significance test for variability,
D
whereas the I2 is an effect size measure for variability and can, therefore, be compared across
TE
analyses. As a heuristic, I2 values of 25%, 50%, and 75% can be considered low, moderate, and
P
was the single extreme value and accounted for more than 50% of the total variance (Q), and the
overall variability (Q) was significant. When outliers were identified, results are presented both
with and without the outlier, with the main interpretation focusing on the findings with the
outlier removed. The exception is that if an analysis of three studies identified one study as an
outlier, it was not removed; with so few studies, identifying outliers becomes unstable.
which continuous moderator variables influenced the magnitude of group differences whereas
the between-level Q statistic was used for categorical moderator variables. The overall Q statistic
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 19
was partitioned into variability across samples that could be explained by the moderator
variability within each level of the moderator (within-level variability, which will be referred to
PT
as Q). A significant between-level Q statistic indicates that the moderator variable explained a
RI
significant portion of the variability across samples. The Q statistic is distributed as a Chi-square,
SC
moderators required at least 2 studies in each level of the moderator whereas meta-regressions
NU
were conducted when there were at least 3 studies reporting on the moderator with sufficient
Demographic Characteristics
D
TE
key demographic and childhood history variables. Extrafamilial offenders were found to be
P
younger (d = 0.21), had higher unemployment rates (d = 0.19), and had lower intelligence than
CE
rates and scoring higher on intelligence tests compared to extrafamilial offenders, intrafamilial
offenders were found to have lower education attainment (d = -0.13). We also restricted this
analysis to the eight studies that reported on both education and intelligence test scores and
found similar findings: Intrafamilial offenders scored higher on intelligence tests compared to
extrafamilial offenders (dfixed and random-effects = -0.20, 95% CI [-0.32, -0.09], Q = 5.92, p = .549, I2
= 0%, N = 1,430), despite having lower educational attainment (dfixed = 0.20, 95% CI [0.04,
0.34], drandom = 0.17, 95% CI [-0.02, 0.35], Q = 11.25, p = .128, I2 = 37.8%%, N = 1,013).
Childhood Difficulties
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 20
Consistent with our hypothesis, intrafamilial offenders were found to have significantly
greater childhood difficulties than extrafamilial offenders (see Table 2). Specifically,
intrafamilial offenders were more likely to have a history of child sexual abuse (d = -0.11),
PT
family abuse (d = -0.31), family neglect (d = -0.25), and poor childhood attachment with parents
RI
(d = -0.24), particularly with mothers (d = -0.24).
SC
Antisocial Tendencies
NU
magnitude but were in the hypothesized direction (see Table 4). Extrafamilial offenders had
MA
greater self-regulation problems (d = 0.10), greater negative peer groups (d = 0.17), scored
higher on measures of antisocial tendencies (d = 0.17), were more likely to have offense-
D
supportive attitudes and beliefs (d = 0.12), and had a larger number of prior offenses (d = 0.20)
TE
than intrafamilial offenders. Perhaps unsurprisingly, given that extrafamilial offenders had
greater criminal history than intrafamilial offenders, differences on PCL-R scores were larger for
P
CE
the Factor 2 (behavioral) subscale (d = 0.26) than the Factor 1 (interpersonal/affective) subscale
(d = 0.10). Extrafamilial offenders were also found to have lower victim empathy (d = 1.05) and
AC
offenders (see Table 3). Extrafamilial offenders had greater sexual interest in children (d = 0.41
for pedohebephilia) and were also more likely to have paraphilias other than pedohebephilia (d =
0.49). Extrafamilial offenders were also more likely to have sexual self-regulation problems
emotional identify with children (d = 0.39), to have greater denial of their sexual offenses (d =
0.27), and to minimize their sexual offenses (d = 0.12) compared to intrafamilial offenders.
PT
Table 5 presents differences between extrafamilial and intrafamilial offenders on other
RI
psychological variables of interest. Extrafamilial offenders tended to have greater problems in
SC
the social sphere (d = 0.12), were lonelier (d = 0.16), and were less likely to have been married
(d = 0.89) or to have cohabitated with a romantic partner (d = 1.16) than intrafamilial offenders.
NU
Groups were similar in psychopathology, with the exception of extrafamilial offenders being
MA
more likely to have been diagnosed or score higher than intrafamilial offenders on measures of
personality disorders (d = 0.11), and intrafamilial offenders more likely to score higher on
D
or repression of feelings.
P
Moderator Analyses
CE
Atypical sexual interests. The average number of victims and the proportion of boy
victims in the intrafamilial and extrafamilial groups were examined as possible moderators for
AC
group differences in any paraphilia, pedohebephilia, or other paraphilias. The average number of
extrafamilial victims was found to moderate group differences on any paraphilia; a greater
number of victims in the extrafamilial group was related to larger effect sizes (Z = 2.04, p = .041,
N = 1,652). The proportion of variability once the meta-analysis was restricted to samples that
reported on the moderator was small in magnitude, ranging from 0 to 38.8% (Mdn I2 = 5%) and,
was not found to moderate group difference on measure of antisocial tendencies (Z = 0.31, p =
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 22
PT
Publication bias. Moderator analyses comparing published to unpublished studies was
RI
conducted for each variable with statistically significant variability as indexed by the Q statistics
(meaning there was significantly more variability between studies that would be expected by
SC
chance alone) and at least two studies in each category. Seven of the 28 analyses reached
NU
statistical significance, meaning that the status of publication (published vs. unpublished)
examined publication bias by calculating Egger‘s regression intercepts to examine the extent to
D
which effect sizes were influenced by ‗missing‘ studies due to publication bias. Only 1 of these 7
TE
variables reached statistical significance, likely given the large number of unpublished studies in
P
the current meta-analysis. Egger‘s intercept was 1.214 (p = .044) for prior sexual offenses. To
CE
help account for this publication bias, the Duvall and Tweedie Trim-and-Fill method was
AC
conducted and adjusted for 9 missing studies to the left of the mean effect size (representing the
missing studies with small sample sizes and lower effect sizes than the mean). The Trim-and-Fill
method resulted in a slightly reduced effect size from that reported in Table 3 (Trim-and-Fill
dfixed = 0.371, 95% CI [0.308, 0.434], Trim-and-Fill drandom = 0.370, 95% CI [0.246, 0.493], Q =
134.36).
These analyses were conducted on a greater number of variables than publication bias analyses
(30 vs. 28) given that time trend analyses only required 3 studies. As seen in Table 6, 10 out of
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 23
30 effect sizes was found to be significantly moderated by study year, with group differences
Discussion
PT
Summary
RI
Our meta-analysis confirmed past work indicating intrafamilial offenders are lower in
SC
antisocial tendencies and atypical sexual interests than extrafamilial offenders (e.g., Seto, 2008).
Unexpectedly, group differences were smaller for antisocial tendencies than atypical sexual
NU
interests, but both were in the same direction. Significant differences in antisocial tendencies
included prior criminal history (but not juvenile delinquency), impulsivity, low victim empathy,
MA
and psychopathy scores. Significant atypical sexual interest differences included the presence of
any paraphilias (especially pedophilia or hebephilia), sexual interest in boys, and sexual self-
D
TE
regulation problems.
higher risk of recidivism on the Static-99 or Static-99R, with a large effect size (d = 1.39). The
CE
Static-99/R is the most widely used actuarial risk assessment for sexual offenders for the
AC
prediction of recidivism. Scores on the Static-99 significantly predict sexual and violent
(including sexual) recidivism (Hanson & Morton-Bourgeon, 2004). This tool contains mostly
characteristics (e.g., prior non-sexual violence, stranger victims). One of the Static-99/R items
pertains to having unrelated child victims. Although this item necessarily increases scores for
extrafamilial offenders, it is not sufficient to explain the large difference in Static-99/R scores
Intrafamilial offenders were less, rather than more, likely to espouse offense-supportive
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 24
attitudes and beliefs (which can also be viewed as an indicator of antisocial tendencies). This is
consistent with the general pattern that intrafamilial offenders are lower risk to reoffend, and
inconsistent with theories that include this factor in explanations for why some men offend
PT
against related rather than unrelated children.
RI
As expected, we found a similar result as McPhail et al. (2013) showing intrafamilial
offenders were less likely to be emotionally congruent with children than extrafamilial offenders,
SC
with a smaller set of studies. Emotional congruence with children is a complex construct. Some
NU
items on measures of emotional congruence with children could reflect healthy parental
attachment (e.g., “I have loved a child at first sight”; Beckett, 1987) and others could be an
MA
indicator of emotional or social immaturity (e.g., “Children remind me of myself”; Beckett,
1987; “When I am with children, I feel like I am one of them”; Wilson, 1999). Fathers who score
D
high on emotional congruence with children because they have a healthy attachment to their
TE
child would be expected to be at lower risk to sexually offend against their child. Conversely,
P
fathers who score high on emotional congruence with children because they emotionally identify
CE
with children or are emotionally or socially immature would be expected to be at higher risk to
AC
sexually offend against their child. Overall, emotional congruence with children has been found
to be positively correlated with atypical sexual interests (McPhail, Hermann, & Fernandez,
2014).
congruence with children and less victim empathy than intrafamilial offenders. Whereas
emotional congruence with children reflects, at least in part, emotional or social immaturity,
deficits in victim empathy have been described as reflecting cognitive distortion, whereby
negative self-appraisals are avoided by changing offenders‘ perceptions of victim impact (e.g.,
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 25
Marshall, Hamilton, & Fernandez, 2001; Marshall, Hudson, Jones, & Fernandez, 1995). In
support of this view, extrafamilial offenders were also found to have greater denial of their
sexual offenses (d = 0.27) and to minimize their sexual offenses more (d = 0.12) than
PT
intrafamilial offenders. Future studies should examine the extent to which intrafamilial offenders
and extrafamilial offenders differ on emotional congruence with children and victim empathy,
RI
with a particular focus on how this construct is operationalized and whether offenders are
SC
biological or sociolegal fathers.
NU
Demographically, extrafamilial offenders were younger, committed their first sexual
offense at a younger age, and were less likely to cohabitate with a partner or be married; this last
MA
result is at least partly an artefact of how we defined the groups. Many intrafamilial offenders
would be genetic or sociolegal fathers to their victims and, thus, it is not surprising they would
D
TE
higher in intelligence than extrafamilial offenders, but despite this small difference, they were
P
Unexpectedly, the largest demographic difference was in sexual orientation, with more
AC
five studies that contributed to this effect size used self-reported information. Three studies
referred specifically to sexual orientation in terms of adults, one referred only to homosexual
activity, and the fifth did not specify. It is possible that the large difference is an artefact of group
classification (i.e., most incest offenders were fathers and thus likely to be in an opposite-sex
marriage or cohabitation relationship). In support for this view, incest offenders were much more
likely to have been married (Mdn = 88%, range from 48 to 100%) than extrafamilial offenders
(Mdn = 56%, range from 36 to 68%). Of note, most studies were completed before same-sex
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 26
marriages became legally recognized in Canada and a growing number of American states.
Consistent with explanations that emphasize abuse history and family-of-origin problems
as precursors of intrafamilial child sexual offending, intrafamilial offenders were indeed more
PT
likely to have histories of childhood sexual abuse, family abuse and neglect, and poor parental
RI
attachment (the comparison specifically on father-child attachment was not significant but was in
the same direction and of similar magnitude to mother-child attachment). There were an
SC
insufficient number of studies to examine spousal and other relationships in the incestuous
NU
family, however, and these relationships might be key to the origins of intrafamilial child sexual
abuse, as we discuss later (Herman, 2012; Maddock & Larson, 1995; Russell, 1986).
MA
The effect sizes were larger for family abuse, neglect, or poor attachment than for sexual
abuse, which might be committed by a family member but might also be committed by unrelated
D
TE
perpetrators. This finding suggests family dysfunction may play a more important role than
sexual abuse in the etiology of intrafamilial sexual offending, and emphasizes the importance of
P
future studies focusing on family dynamics. It also suggests, as we briefly discuss later, that
CE
some treatment foci might differ for intrafamilial compared to extrafamilial offenders.
AC
Contrary to the predictions derived from some clinical accounts of incest, intrafamilial
offenders were significantly lower on any social problems and loneliness, and there were few
differences in psychopathology. Intrafamilial offenders did score higher in repression and lower
in personality disorder, but there were no significant group differences in anxiety, depression,
Motivation-Facilitation Model
tendencies and atypical sexual interests – do not explain why some men sexually offend against
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 27
related children instead of unrelated children. This does not mean these factors are not relevant in
explanations of intrafamilial child sexual offending; even if intrafamilial offenders are relatively
low in these risk dimensions for sexual offending, they may be higher than other men. For
PT
example, intrafamilial offenders score higher on attitudes supportive of child sexual abuse
RI
compared to nonoffenders (d = .20), non-sex offenders (d = .64), and sex offenders against adults
(d = .64; Hayashino, Wurtele, & Klebe, 1995). Similarly, Nexhipi (1992) found that offenders
SC
with child victims more generally scored higher on these offense-supportive attitudes than
NU
nonoffenders or other kinds of offenders. Beech et al. (2008) found that sex offenders with child
victims, whether related or unrelated, performed differently than non-sex offenders on a task
MA
comparing attention to pictures of children to pictures of animals. At the same time, both Mills,
Anderson, and Kroner (2004) and Nexhipi (1992) found that intrafamilial offenders scored lower
D
TE
than non-sex offenders on measures of general antisocial attitudes and beliefs, again supporting
the idea that intrafamilial offenders are relatively low in (some) antisocial tendencies. This is
P
also consistent with the results reported by Porter et al. (2000), who found that both intrafamilial
CE
and extrafamilial offenders had lower scores on psychopathy than sex offenders with adult
AC
victims or non-sex offenders. In short, established factors do not adequately explain victim
choice among sex offenders against children (i.e., intrafamilial vs. extrafamilial offenders).
These factors, however, are still useful in explaining offender group differences and sexual
Family Dynamics
these were the data available in existing comparison studies. Yet a great deal of clinical and
theoretical literature has focused on the role played by family dynamics, including spousal
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 28
studies compared intrafamilial and extrafamilial offenders in their early involvement in their
PT
child‘s life, no studies examined parent-child similarities, no studies examined suspected or
actual infidelity of the offender‘s spouse/partner, and only one study examined parental absence.
RI
This is unfortunate given that family-level factors are a promising candidate for explaining
SC
intrafamilial child sexual offending.
NU
Given that intrafamilial offenders are relatively low on antisocial tendencies and atypical
sexual interests, family factors could explain why some sex offenders against children offend
MA
against related children and others against unrelated children. For example, one promising factor
is a failure to develop solicitous parenting, which could be the result of the offender‘s own poor
D
TE
experiences as a child, patriarchical attitudes and beliefs regarding the roles of fathers, mothers,a
nd children, or a belief that a putatively genetically related child is not in fact related (see Seto,
P
2008). The following sections discuss potential family-level causal candidates as a call for more
CE
Spousal relationship quality. Sociological and psychological theories about incest have
suggested that dysfunctional family relationships play a major role in incestuous sexual
offending (e.g., Herman, 2012; Maddock & Larson, 1995; Russell, 1986). In these family-
focused theories, intrafamilial child sexual abuse might occur when the parental relationship has
broken down and the mother is sexually and emotionally unavailable; for example, a father may
turn to his eldest daughter to fulfill his sexual and emotional needs because his spouse is
depressed, disinterested in having sex with him, or are not getting along. In essence, the daughter
There are some data consistent with this view. Lang, Langevin, Van Santen, Billingsley
and Wright (1990) compared 92 intrafamilial offenders (86% were genetic fathers or step-
fathers) with 42 nonoffending controls and found that the intrafamilial offenders reported less
PT
communication with their partner, felt more lonely, and were less satisfied with their partner.
RI
The intrafamilial offenders and nonoffending controls did not differ in the length of their
marriages or their number of prior marriages. Unfortunately, Lang et al. did not include a
SC
comparison group of non-intrafamilial sex offenders. As such, it is unknown the extent to which
NU
differences between groups were due to differences in criminality rather than difference in
propensity to commit intrafamilial sexual offenses; it might be the case, for example, that
MA
individuals who engage in criminal behavior have more relationship problems, on average, than
those who have not engaged in criminal behavior. In a retrospective survey of 2,304 women,
D
TE
and negatively related to maternal affection (Stroebel et al., 2012). Over half of parents in
P
incestuous families report sexual discord (Saunders, Lipovsky, & Hanson, 1995). Dadds, Smith,
CE
Webber, and Robinson (1991) found incestuous families were more troubled – reporting higher
AC
conflict, lower emotional expression, and lower cohesion – than non-incestuous families.
Relationship with child victims. Daly and Wilson (1998) suggested lower parental
solicitude increases the risk of maltreatment, including sexual and physical abuse. Individuals
who have poorer relationships with related children are more likely to sexually abuse them. In
support to this hypothesis, victims of intrafamilial sexual offending describe the offending father
hypothesized that early physical propinquity induces sexual indifference among children
growing up together and, as such, not growing up together increases the risk for incest.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 30
been reported to have lower involvement in parental care (Herman, 2012; Parker & Parker, 1986;
Williams & Finkelhor, 1995). In a survey of college students, Bevc and Silverman (1993) found
PT
that respondents who admitted attempting to have intercourse or had intercourse with a sibling
RI
were more likely to have been separated from that sibling for more than a year during their first
six years of life than those who denied any sexual intercourse. This difference was not found
SC
when comparing respondents who engaged in non-reproductive sexual behaviors such as kissing
NU
and fondling. The group differences were replicated by Bevc and Silverman (2000) in a mixed
offenders and their related child victims, before the incestuous offenses occurred, compared to
D
TE
the quality of the relationships the intrafamilial offender had with other related but non-
victimized children.
P
Family dysfunction. Descriptive studies have cited a variety of family factors as relevant
CE
to intrafamilial sexual abuse, including poor sexual boundaries, lack of cohesion, rigid
AC
organization, enmeshment and poor parental care/supervision (e.g., Smith & Israel, 1987).
Unfortunately, these clinically informed studies have not included suitable non-intrafamilial
offender comparison groups and psychometrically sound measures of these family factors.
Nonetheless, consistent with these speculations, O‘Brien (1991) found that adolescent sibling
intrafamilial offenders were from more disturbed families than adolescents who had sexually
adolescent sibling intrafamilial offenders and 28 non-sibling offenders and found the former
scored higher on parental rejection, parental discord, negative family atmosphere, and family
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 31
Kin recognition. Theories of incest suggest that incestuous behaviours are more
likely to occur when kinship cues are absent (e.g., Westermarck, 1891/1921). Seto (2008)
PT
speculated that some biological father offenders may in fact behave like sociolegal fathers
RI
because they do not believe (consciously or unconsciously) that the child is genetically related.
Cues might include absences around the time of conception, suspicion of the mother‘s infidelity,
SC
or perceived non-resemblance in physical appearance or personality traits. Some research has
NU
examined the role of kin recognition mechanisms in incest aversion or avoidance towards parents
or towards siblings (Antfolk, Lindqvist, Albrecht, & Santtila, 2014; Lieberman, Tooby &
MA
Cosmides, 2007).
It is possible that the link between family problems and incestuous behaviour is directly
D
explained by a third factor: weak kinship cues (e.g., the father‘s belief that the child is not
TE
related, frequent absences around the time of conception). Weak kinship cues would be expected
P
to be associated with poor family dynamics and, as such, can explain why poor family dynamics
CE
are more common amongst incestuous families. Further studies examining these variables would
AC
Clinical Implications
Findings from this meta-analysis have implications for clinical practices because it
identifies domains in which intrafamilial offenders are more or less likely to be problematic.
Currently, sex offender treatment programs typically focus on risk-related domains such as
paraphilic sexual interests (McGrath, Cumming, Burchard, Zeoli, & Ellerby, 2010). Yet, the
current meta-analysis found that intrafamilial offenders are generally less problematic in these
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 32
domains than other sex offenders against children, though they are still more likely to have
problems in the domains surveyed here than the general population (Beech et al., 2008;
Hayashino et al., 1995; Nexhipi, 1992). Our findings highlight that sex offender treatment
PT
programs may benefit from being tailored for intrafamilial versus extrafamilial offenders by
RI
providing additional content and differences in emphasis. For example, intrafamilial offenders
were more likely to have family problems, including histories of maltreatment and poor parent-
SC
chlid attachments. This suggests that treatment targeting family functioning might be valuable
NU
for intrafamilial offenders. It also suggests that targeting at-risk families for prevention efforts
could be fruitful for intrafamilial sexual abuse as well as other negative outcomes.
MA
Future research is needed to examine the extent to which such treatment effort are
effective in reducing intrafamilial sexual abuse and other forms of family abuse. Surprisingly
D
TE
little research comparing intrafamilial and other sex offenders with children has examined family
dynamics, despite the theoretical and clinical importance of this domain. A better understanding
P
of family dynamics is needed to inform prevention and family-based treatment for intrafamilial
CE
Limitations
Although intrafamilial sexual offending is common and recognized as a problem that cuts
across health, social service, and justice systems, the correlates of intrafamilial sexual offending
are not well-understood. Part of the challenge in discovering the correlates of intrafamilial sexual
offending is definitional: What relationships and behaviors comprise incest? There is consensus
that sexual contacts involving members of a genetic nuclear family would constitute incest, but
at what degree of genetic relatedness would sexual activity no longer be considered incestuous?
What about sociolegal relationships? These distinctions matter because there may be important
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 33
fathers (e.g., step-fathers, adoptive fathers). Incest avoidance mechanisms that are driven by the
PT
combining harmful recessive alleles when closely related individuals mate and produce offspring
– would only apply to genetic fathers, not to those who take on a father role through
RI
cohabitation, remarriage, or adoption. Relatedly, step-fathers are disproportionately more likely
SC
to perpetrate sexual abuse than genetic fathers (Daly & Wilson, 1998; Sariola & Uutela, 1996).
NU
Some studies have shown that these two groups of incestuous fathers do not differ in their
sexual response to stimuli depicting children (e.g., Rice & Harris, 2002; Seto et al., 1999).
MA
However, studies typically combine intrafamilial offenders with different relationships to their
victims, and explanations for intrafamilial sexual offending may depend on offender-victim
D
TE
relationship. For example, spousal relationship quality is more relevant for genetic fathers and
sociolegal fathers than for siblings (who typically are not much older than child victims and thus
P
unlikely to have a spouse). As a second example, intrafamilial offenders may differ in the
CE
likelihood of having pedophilia, as suggested by Seto and colleagues (1999), who found
AC
biological father offenders did not significantly differ from step-father offenders but did score
lower in their sexual arousal to children than extended family member offenders such as uncles
or grandfathers.
The costs of inbreeding and the strength of incest taboos vary by relationship, and so it is
possible that different explanations apply for sexual abuse by nuclear family members compared
to extended family members, and for sexual abuse by genetic relationships compared to
sociolegal relationships. Step-fathers or adoptive fathers, for example, are not typically present
from the child victim‘s birth on, whereas genetic fathers typically are. Westermarck‘s
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 34
(1891/1921) hypothesis about early propinquity leading to sexual indifference or aversion would
suggest differences between genetic and sociolegal fathers as a result (see Seto, 2008). We need
PT
intrafamilial sexual offending by relationship.
RI
An important limitation is that studies included in this meta-analysis compared
SC
include non-fathers. A more refined analysis would compare extrafamilial offenders with
NU
children of their own to incestuous fathers. Such a comparison would be able to answer why
some fathers decide to offend against a related child, while other decided to offend against an
MA
unrelated child despite having access to children of their own.
Other group comparisons could have been made, but each has its limitations. For example,
criminal status, but does not address the confound that one group has engaged in illegal sexual
CE
behavior whereas the other group has engaged in illegal nonsexual behavior. Nonsexual
AC
offenders would also not be an ideal comparison group because any difference found between
propensity to commit a sexual offense rather than the propensity to commit incest. Nonoffending
men are not an ideal comparison group because any differences found between nonoffending
men and incest offenders could be attributed to differences in criminality more generally rather
than any propensity to commit incest, because the latter group have committed an illegal act.
There are also likely to be many sociodemographic differences, including education level,
operational definitions to define groups and to assess variables of interest. Intrafamilial offender
groups not only varied in their composition in terms of relationship to victim, but studies used
PT
different selection criteria. For example, one study might include live-in boyfriends who have
RI
cohabitated for six months (this would meet the common-law definition in Ontario, Canada)
whereas another study might require a minimum of one year of cohabitation. In many studies,
SC
extrafamilial offenders could also have related child victims, which might attenuate group
NU
differences compared to study designs that compared intrafamilial-only offenders with
operational definitions of the variables of interest within studies (van den Noortgate, López-
D
TE
López, Marín-Martínez, & Sánchez-Meca, 2013). Such an approach could provide a more
detailed examination of the findings in the current meta-analysis (e.g., exploring differences
P
However, given the broad inclusion criteria (i.e., any study comparing intrafamilial and
interests, and other psychological variables), it is possible that some studies were missed. Meta-
analyses are influenced to the extent relevant studies are unduly versus randomly excluded.
Importantly, we found that for some variables, published studies had larger effect sizes than
those derived from unpublished studies, suggesting a publication bias toward statistically
significant findings. Given that our search strategy identified a large number of unpublished
studies, the only variable that was found to be missing studies due to publication bias (as
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 36
assessed by Egger‘s regression intercepts) was prior sexual offenses; we reported an adjusted
mean effect size using the Trim and Fill method. Relatedly, meta-regression analysis suggested
that average effect size (group differences) was influenced by study year for some comparisons,
PT
which may reflection publication bias or sample selection biases over time.
RI
Another common limitation in criminal justice research is that offenders were usually
grouped on the basis of official charges or convictions. As such, undetected offenses against
SC
related or unrelated children were not taken into account, which would probably serve to
NU
attenuate group differences that exist because some intrafamilial offenders may have unrelated
victims and some extrafamilial offenders may also have offended against related children.
MA
The studies analyzed here were comprised of clinical or forensic samples of identified
offenders. The results that were obtained may not generalize to the broader population of
D
TE
individuals who have committed intrafamilial or extrafamilial child sexual abuse, given many
such individuals are not reported or detected by authorities. In addition, all of the studies
P
groups after the offenses had occurred. This means self-reports of interpersonal and family
AC
functioning might be influenced by recall or other biases. For example, the group difference in
loneliness might reflect, at least in part, the effects of being identified as a sex offender, rather
than an emotional precursor that might explain the commission of sexual offenses. The goal of
this meta-analysis was to identify potential correlates that could then be examined more closely
in longitudinal studies, such as following at-risk families, and if scientifically and ethically
Future Directions
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 37
We have already noted a number of directions for future research, including studies
comparing different subtypes of intrafamilial offenders, and longitudinal and possibly quasi-
experimental studies to further examine causal candidates identified in the current meta-analysis.
PT
The significant differences between intrafamilial and extrafamilial offenders on a number of
RI
demographic characteristics such as age and ethnicity suggest further research is needed to
determine if these characteristics moderate group differences or moderate the roles of other
SC
variables of interest. There is also a need for large sample studies using multiple measures to test
NU
more complex models of intrafamilial sexual offending. The comparisons meta-analyzed in the
current study highlight potential correlates that may increase the likelihood of intrafamilial rather
MA
than extrafamilial offending. This kind of study design cannot identify causal relationships,
complex relationship involving mediators or interaction effects, or the extent that the variable is
D
TE
useful in explaining the occurrence of child sexual offending in general. Given the prevalence
and costs associated with criminal offending, a renewed focus on intrafamilial sexual offending
P
Acknowlegments
Thank you to Marcus Boccaccini, R. Karl Hanson, Yolanda Fernandez, Genevieve Martin, Jill
Levenson, and Wineke Smid who took the time to share unpublished data with us and/or
PT
answered our questions. We are also grateful to Karl Hanson and Martin Lalumière for their
RI
helpful comments on an earlier version of this paper.
SC
NU
MA
D
P TE
CE
AC
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 39
References
Abel, G. G., Becker, J. V., Murphy, W. D., & Flanagan, B. (1981). Identifying dangerous child
PT
management and treatment (pp. 116-137). New York, NY: Brunner/Mazel.
RI
Antfolk, J., Karlsson, M., Bäckström, A., & Santtila, P. (2012). Disgust elicited by third-party
SC
Evolution and Human Behavior, 33, 217-223. doi:10.1016/j.evolhumbehav.2011.09.005
NU
Antfolk, J., Lindqvist, H., Albrecht, A., & Santtilla, P. (2013). Self-reported availability of
Beech, A. R., Kalmus, E., Tipper, S. P., Baudouin, J., Flak, V., & Humphreys, G. W. (2008).
D
Beckett, R. C. (1987). Children and sex questionnaire. Oxford, United Kingdom: Oxford
CE
Forensic Services.
AC
Beitchman, J. H., Zucker, K. J., Hood, J. E., DaCosta, G. A., Akman, D., & Cassavia, E. (1992).
A review of the long-term effects of child sexual abuse. Child Abuse & Neglect, 16, 101-
118. doi:10.1016/0145-2134(92)90011-F
Bevc, I., & Silverman, I. (1993). Early proximity and intimacy between siblings and incestuous
behavior: A test of the Westermarck theory. Ethology and Sociobiology, 14, 171-181.
doi: 10.1016/0162-3095(93)90004-2
Bevc, I., & Silverman, I. (2000). Early separation and sibling incest: A test of the revised
Westermarck theory. Evolution and Human Behavior, 21, 151-161. doi: 10.1016/S1090-
5138(99)00041-0
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 40
Bogaerts, S., Vanheule, S., & Declercq, F. (2005). Recalled parental bonding, adult attachment
style, and personality disorders in child molesters: a comparative study. The Journal of
PT
Bonomi, A. E., Anderson, M. L., Rivara, F. P., Cannon, E. A., Reid, R. J., & Thompson, R. S.
RI
(2008). Health care utilization and costs associated with childhood abuse. Journal of
SC
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to
NU
meta-analysis. Chichester, West Sussex, U.K.: Wiley.
Chaplin, T. C., Rice, M. E., & Harris, G. T. (1995). Salient victim suffering and the sexual
MA
responses of child molesters. Journal of Consulting and Clinical Psychology, 63, 249-
Chen, L. P., Murad, M. H., Paras, M. L., Colbenson, K. M., Sattler, A. L., Goranson, E. N., ...
systematic review and meta-analysis. Mayo Clinic Proceedings, 85, 618-629. doi:
CE
10.4065/mcp.2009.0583
AC
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. New York: Routledge.
Dadds, M., Smith, M., Webber, Y., & Robinson, A. (1991). An exploration of family and
individual profiles following father-daughter incest. Child Abuse & Neglect, 15, 575-586.
doi: 10.1016/0145-2134(91)90041-B
Daly, M., & Wilson, M. (1998). The truth about Cinderella: A Darwinian view of parental love.
Finkelhor, D. (1984). Child sexual abuse: New theory and research. New York, NY: Free Press.
Finkelhor, D. (1994). The international epidemiology of child sexual abuse. Child Abuse &
PT
Firestone, P., Dixon, K. L., Nunes, K. L., & Bradford, J. M. (2005). A comparison of incest
RI
offenders based on victim age. Journal of the American Academy of Psychiatry and the
SC
Fisher, D., Beech, A. R., & Browne, K. (1999). Comparison of sex offenders to nonoffenders on
NU
selected psychological measures. International Journal of Offender Therapy and
(Eds.), The handbook of research synthesis (pp. 245-260). New York: Russell Sage
D
TE
Foundation.
Frenzel, R. R., & Lang, R. A. (1989). Identifying sexual preferences in intrafamilial and
P
extrafamilial child sexual abusers. Annals of Sex Research, 2, 255-275. doi: 10.1177/
CE
10790632 8900200304
AC
Freund, K., Watson, R., & Dickey, R. (1991). Sex offenses against female children perpetrated
by men who are not pedophiles. The Journal of Sex Research, 28, 409-423. . doi: 10.2307
/3812711
Garber, J., & Hollon, S. D. (1991). What can specificity designs say about causality in
2909.110.1.129
Gibbens, T. C., Soothill, K. L., & Way, C. K. (1978). Sibling and parent-child incest offenders:
Hall, G. C. N., & Hirschman, R. (1992). Sexual aggression against children: A conceptual
perspective of etiology. Criminal Justice and Behavior, 19, 8-23. doi: 10.1177/
0093854892019001003
PT
Hanson, R. K., & Bussière, M. T. (1998). Predicting relapse: a meta-analysis of sexual offender
RI
recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362. doi:
10.1037/0022-006X.66.2.348
SC
Hanson, R. K., Gizzarelli, R. & Scott, H. (1994). The attitudes of incest offenders: Sexual
NU
entitlement and acceptance of sex with children. Criminal Justice and Behavior, 21, 187-
Hanson, R. K., & Morton-Bourgon, K. E. (2009). The accuracy of recidivism risk assessments
P
Hanson, R. K., Harris, A. J., Scott, T. L., & Helmus, L. (2007). Assessing the risk of sexual
offenders on community supervision: The Dynamic Supervision Project (Vol. 5). Ottawa,
Hartley, C. C. (1998). How incest offenders overcome internal inhibitions through the use of
cognitions and cognitive distortions. Journal of Interpersonal Violence, 13, 25-39. doi:
10.1177/088626098013001002
Hasselblad, V., & Hedges, L.V. (1995). Meta-analysis of screening and diagnostic tests.
Hawes, S. W., Boccaccini, M. T., & Murrie, D. C. (2013). Psychopathy and the combination of
PT
243. doi: 10.1037/a0030391
RI
Hayashino, D.S., Wurtele, S.K., & Klebe, K.J. (1995). Child molesters: An examination of
SC
095010001007
NU
Herman, J. L. (2012). Father-daughter incest. Cambridge, MA: Harvard University Press.
Higgins, J., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in
MA
meta-analyses. British Medical Journal, 327, 557-560. doi:10.1136/bmj.327.7414.557
Jespersen, A. F., Lalumière, M. L., & Seto, M. C. (2009). Sexual abuse history among adult sex
D
TE
offenders and non-sex offenders: A meta-analysis. Child Abuse and Neglect, 33, 179-192.
doi: 10.1016/j.chiabu.2008.07.004
P
offenders as a function of victim age. Archives of Sexual Behavior, 20, 187-197. doi:
AC
10.1007/BF01541943
Koenig, W. D., Haydock, J., & Stanback, M. T. (1998). Reproductive roles in the cooperatively
Lang, R. A., Black, E. L., Frenzel, R. R., & Checkley, K. L. (1988). Aggression and erotic
attraction toward children in incestuous and pedophilic men. Annals of Sex Research, 1,
Lang, R. A., Langevin, R., Van Santen, V., Billingsley, D., & Wright, P. (1990). Marital
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 44
relations in incest offenders. Journal of Sex and Marital Therapy, 16, 214-229. doi:
10.1080/00926239008405459
Leavitt, G. C. (2005). Incest and inbreeding avoidance: a critique of Darwinian social science.
PT
Lewiston, NY: Edwin Mellen Press.
RI
Lieberman, D., Tooby, J., & Cosmides, L. (2007). The architecture of human kin detection.
SC
Lipsey, M. W. (2003). Those confounded moderators in meta-analysis: Good, bad, and ugly. The
NU
Annals of the American Academy of Political and Social Science, 587, 69-81. doi:
10.1177/0002716202250791
MA
Lu, Y. C., & Lung, F.-W. (2012). Perceived parental attachment, personality characteristics, and
Marshall, W. L., Hamilton, K., & Fernandez, Y. (2001). Empathy deficits and cognitive
AC
distortions in child molesters. Sexual Abuse: A Journal of Research and Treatment, 13,
123-130. doi:10.1177/107906320101300205
Marshall, W. L., Hudson, S. M., Jones, R., & Fernandez, Y. M. (1995). Empathy in sex
McKillop, N., Smallbone, S., Wortley, R., & Andjic, I. (2012). Offenders‘ attachment and sexual
abuse onset a test of theoretical propositions. Sexual Abuse: A Journal of Research and
McGrath, R. J., Cumming, G. F., Burchard, B. L., Zeoli, S., & Ellerby, L. (2009). Current
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 45
practices and emerging trends in sexual abuser management. Brandon, Vermont: The
Current_Practices_Emerging_Trends.pdf
PT
McPhail, I. V., Hermann, C. A., & Fernandez, Y. M. (2014). Correlates of emotional congruence
RI
incarcerated sample. Child Abuse & Neglect, 38, 336-346. doi:
SC
10.1016/j.chiabu.2013.10.002
NU
McPhail, I. V., Hermann, C. A., & Nunes, K. L. (2013). Emotional congruence with children and
Mills, J. F., Anderson, D., & Kroner, D. G. (2004). The antisocial attitudes and associates of sex
D
TE
offenders. Criminal Behaviour and Mental Health, 14, 134-145. doi: 10.1177/
0093854804268755
P
Nexhipi, G. (1992). A structural analysis of the attitudes of those who sexually abuse children:
CE
abuse: Frontline research and evaluation (pp. 75-92). Newbury Park, CA: Sage.
Paras, M. L., Murad, M. H., Chen, L. P., Goranson, E. N., Sattler, A. L., Colbenson, K. M., ...
Zirakzadeh, A. (2009). Sexual abuse and lifetime diagnosis of somatic disorders. Journal
Parker, H., & Parker, S. (1986). Father-daughter sexual abuse: An emerging perspective.
0025.1986.tb03486.x
Pemberton, A. E., & Wakeling, H. C. (2009). Entitled to sex: Attitudes of sexual offenders.
PT
Pereda, N., Guilera, G., Forns, M., & Gómez-Benito, J. (2009). The prevalence of child sexual
RI
abuse in community and student samples: A meta-analysis. Clinical Psychology Review,
SC
Porter, S., Fairweather, D., Drugge, J., Herve, H., Birt, A., & Boer, D. P. (2000). Profiles of
NU
psychopathy in incarcerated sexual offenders. Criminal Justice and Behavior, 27, 216-
7894(79)80057-X
Rice, M. E. & Harris, G. T. (2002). Men who molest their sexually immature daughters: is a
P
10.1037//0021-843X.111.2.329
AC
Criminology.
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. doi:10.1037/0033-2909.124.1.22
Roberts, R., O‘Connor, T., Dunn, J., & Golding, J. (2004). The effects of child sexual abuse in
later family life; mental health, parenting and adjustment of offspring. Child Abuse &
Russell, D. E. (1986). The secret trauma: Incest in the lives of girls and women. New York, NY:
Basic.
Sánchez-Meca, J., Marín-Martínez, F., & Chacón-Moscoso, S (2003). Effect-size indices for
PT
dichotomized outcomes in meta-analysis. Psychological Methods, 8, 448-467. doi:
RI
0.1037/1082-989X.8.4.448
Sariola, H., & Uutela, A. (1996). The prevalence and context of incest abuse in Finland. Child
SC
Abuse & Neglect, 20, 843-850. doi: 10.1016/0145-2134(96)00072-5
NU
Saunders, B. E., Lipovsky, J. A., & Hanson, R. F. (1995). Couple and familial characteristics of
father-child incest families. Journal of Family Social Work, 1, 5-25. doi: 10.1300
MA
/J039v01n02_02
Seto, M. C. (2008). Pedophilia and sexual offending against children: Theory, assessment, and
D
TE
Association.
CE
Seto, M. C., & Barbaree, H. E. (1999). Psychopathy, treatment behavior and sex offender
AC
10.1177/088626099014012001
Seto, M. C., & Lalumière, M. L. (2010). What is so special about male adolescent sexual
Seto, M. C., Lalumière, M. L., & Kuban, M. (1999). The sexual preferences of incest offenders.
Schulze, R. (2007). Current methods for meta-analysis: Approaches, issues, and developments.
3409.215.2.90
PT
Smith, H., & Israel, E. (1987). Sibling incest: A study of the dynamics of 25 cases. Child Abuse
RI
& Neglect, 11, 101-108. doi: 10.1016/0145-2134(87)90038-X
Stroebel, S. S., O'keefe, S. L., Beard, K. W., Kuo, S. Y., Swindell, S. V., & Kommor, M. J.
SC
(2012). Father–daughter Incest: Data from an anonymous computerized survey. Journal
NU
of Child Sexual Abuse, 21, 176-199. doi: 10.1080/10538712.2012.654007
Stoltenborgh, M., van IJzendoorn, M. H., Euser, E. M., & Bakermans-Kranenburg, M. J. (2011).
MA
A global perspective on child sexual abuse: Meta-analysis of prevalence around the
Thornhill, N.W. (1991). An evolutionary analysis of rules regulating human inbreeding and
CE
525X00066449
Underhill, J., Wakeling, H. C., Mann, R. E., & Webster, S. D. (2008). Male sexual offenders'
emotional openness with men and women. Criminal Justice and Behavior, 35, 1156-
Van den Noortgate, W., López-López, J. A., Marín-Martínez, F., & Sánchez-Meca, J. (2013).
Wakeling, H. C., Webster, S., Moulden, H. M., & Marshall, W. L. (2007). Decisions to offend in
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 49
men who sexually abuse their daughters. Journal of Sexual Aggression, 13, 81-99. doi:
10.1080/13552600701521330
Ward, T., Polaschek, D., & Beech, A. R. (2006). Theories of sexual offending. Chicester, UK:
PT
Wiley & Sons.
Westermarck, E. A. (1891/1921). The history of human marriage (5th ed.). London, U.K.:
RI
MacMillan.
SC
Williams, L. M., & Finkelhor, D. (1990). The characteristics of incestuous fathers: A review of
NU
recent studies. In W.L Marshall, D. R. Laws, & H. E. Barbaree (Eds.). Handbook of
sexual assault: Issues, theories and treatment of the offenders (pp. 231-255). New York,
MA
NY: Plenum Press.
Williams, L. M., & Finkelhor, D. (1995). Paternal caregiving and incest: Test of a biosocial
D
TE
Widom, C. S., Czaja, S. J., & Dutton, M. A. (2008). Childhood victimization and lifetime
P
Wilson, R. J. (1999). Emotional congruence in sexual offenders against children. Sexual Abuse:
AC
Wolf, A. P., & Durham, W.H. (Eds.).(2004). Inbreeding, incest, and the incest taboo: the state of
knowledge at the turn of the century. Stanford, CA: Stanford University Press.
functioning when compared to adolescent nonsibling sex offenders. Child Abuse &
Table 1
PT
Study Authors Country Intrafam Extrafam Location Adversarial Treatment
RI
N N Degree Received?
1 Bagley & Prichard (2000) United Kingdom 105 269 Community NA No
SC
2 Barnes (2000) United States 124 73 Community Moderate Mixed
3 Barsetti, Earls, Lalumière, & Belanger (1998) Canada 19 20 Community Unknown Unknown
4 Flak (2011) United Kingdom 12 14 Institution Low No
NU
5 Beggs & Grace (2008) New Zealand 124 92 Institution Moderate Yes
6.1 Langevin, Paitich, Freeman, Mann, & Hardy Canada 230 318 Unknown High No
MA
(1978)
6.2 Stermac, Hall, & Henskens (1989)
6.3 Langevin, Wright, & Handy (1988)
6.4 Freund, Watson, & Dickey (1991)
ED
6.5 Seto, Lalumière, & Kuban (1999)
7 Clounch (2008) United States 140 143 Institution Moderate No
PT
8 Cole (1995) United States 42 29 Community Moderate No
9 Curtin & Niveau (1998) Switzerland 11 34 Community High No
10 de Chabalier & Forzan (2009) France 26 17 Community Moderate Unknown
CE
11 Dennison, Stough & Birgden (2001) Australia 28 32 Institution Unknown Unknown
12.1 Ducro (2009) Belgium 28 26 Combined Moderate No
AC
Table 1 continued.
PT
N N Degree Received?
20.2 Jamieson (1997)
20.3 Simourd & Malcolm (1998)
RI
20.4 Jamieson & Marshall (2000)
21 Kaufman, Harbeck-Weber, & Rudy (1994) United States 16 15 Institution Low Yes
SC
22 Kennedy (1995) United Kingdom 30 10 Institution Low Unknown
23 Kovach (2002) United States 84 132 Institution Moderate No
NU
24 Langton et al. (2008) Canada 17 33 Institution Moderate Yes
25.1 Mann, Webster, Wakeling, & Marshall (2007) United Kingdom 476 473 Institution Moderate No
25.2 Webster, Mann, Thorton, & Wakeling (2007)
MA
25.3 Underhill, Wakeling, Mann, & Webster (2008)
26.1 Marshall, Barbaree, & Eccles (1991) Canada 48 91 Community Moderate No
26.2 Marshall (1988)
ED
26.3 Marshall, Barbaree, & Christophe (1986)
26.4 Marshall & Barbaree (1988)
27.1 Oliver (2004) United States 460 1333 Combined High Unknown
27.2
27.3
Matala (2008)
Boughner (2010) PT
CE
28.1 McCoy (1997) Canada 295 255 Institution High Unknown
28.2 Wexler (2005)
28.3 Firestone, Nunes, Moulden, Broom, & Bradford
AC
(2005)
28.4 Firestone, Bradford, Greenberg, & Serran (2000)
28.5 Pawlak, Boulet, & Bradford (1991)
28.6 Allnut, Bradford, Greenberg, & Curry (1996)
28.7 Greenberg, Bradford, Firestone, & Curry (2000)
29 Medlin (1995) United States 60 63 Institution Low No
30 Mills, Anderson, & Kroner (2004) Canada 27 30 Institution Low Unknown
31 Miner and Dwyer (1997) United States 27 27 Community Moderate No
32.1 Tetreault (2005) Canada 84 104 Community Moderate Mixed
32.2 Muschang (2007)
Table 1 continues.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 52
Table 1 continued.
PT
N N Degree Received?
33 Murphy, Haynes Stalgaitis, & Flanagan (1986) United States 22 38 Community Moderate Yes
34 Nagayama-Hall, Maiuro, Vitaliano, & Proctor United States 58 348 Institution High Unknown
RI
(1986)
35 Nexhipi (1992) Canada 21 24 Institution Low No
SC
36 Norton (1990) United States 95 28 Community Moderate No
37 O'Bannon (1991) United States 13 30 Community Moderate Yes
NU
38 Olver & Wong (2006) Canada 29 25 Institution Moderate Unknown
39 Panton (1979) United States 35 28 Institution Moderate Unknown
40 Pensinger (1995) United States 30 23 Community High Yes
MA
41 Porter et al. (2000) Canada 37 64 Institution Moderate Unknown
42 Proulx, Lussier, Ouimet, & Boutin (2007) Canada 188 115 Institution NA No
43 Quinsey, Chaplin, & Carrigan (1979) Canada 16 16 Institution Unknown Unknown
ED
44 Rada, Kellner, Laws, & Winslow (1978) United States 32 203 Institution Low Yes
45 Rattenbury (1986) United States 16 88 Institution High No
46 Rice & Harris (2002) Canada 45 139 Combined NA Unknown
47
48
Rosenberg, Abell, & Mackie (2005)
Schur (1986) PTUnited States
United States
30
50
81
50
Institution
Unknown
High
High
No
Yesa
CE
49 Simkins, Ward, Bowman, & Rinck (1989) United States 80 39 Community Moderate No
50 Smallbone & Dadds (1998) Australia 16 16 Institution Low Unknown
51.1 Stripe (2003) Canada 19 22 Combined Low Yes
AC
Table 1 continued.
PT
N N Degree Received?
53.6 Lang & Frenzel (1988)
54.1 Valliant & Antonowicz (1992) Canada 15 34 Institution Moderate Mixed
RI
54.2 Valliant & Blasutti (1992)
54.3 Valliant, Gauthier, Pottier, & Kosmyna (2000)
SC
55 Welk (2001) United States 15 15 Community Low Yes
56 Wilson (1999) Canada 62 72 Combined Low Yes
NU
57 Yates & Kingston (2006) Canada 24 19 Institution Moderate No
58 Beech, Fisher, & Beckett (1998) United Kingdom 60 47 Institution Moderate No
59 Gullikson (1993) United States 76 66 Institution Moderate Yes
MA
60 Pittman (1981) United States 15 15 Community Low Yes
61.1 Eher, Rettenberger, & Schilling (2010) Austria 228 201 Institution Moderate No
61.2 Eher & Ross (2006)
ED
61.3 Eher (2006)
62 Schmidt, Gykiere, Vanhoeck, Mann, & Banse Belgium 19 35 Community Moderate No
(2013)
63.1
63.2
Martin & Tardiff (2013)
Martin & Tardiff (2008) PT Canada 55 67 Combined Low Mixed
CE
64 Akca (1986) United States 42 31 Institution Moderate Unknown
65 Beech et al. (2008) United Kingdom 16 19 Institution Low No
66.1 Boccaccini, Murrie, Hawes, Simpler, & Johnson United States 1052 2805 Institution High Unknown
AC
(2010)
66.2 Murrie, Boccaccini, Caperton, & Rufino (2012)
66.3 Boccaccini, Murrie, Caperton, & Hawes (2009)
67 Seto & Fernandez (2011) Canada 85 139 Institution Moderate No
68 Smid, Kamphuis, Weyer, & Van Beek (2013) Netherlands 44 195 Institution High No
69 Choudhry (1995) Canada 12 14 Institution Moderate Yes
70 Blanchard et al. (2006) Canada 92 60 Unknown Unknown No
71 Malcolm, Andrews & Quinsey (1993) Canada 25 36 Institution Moderate No
72 Harris, Smallbone, Dennison, & Knight (2009) United States 34 236 Institution High Mixed
73.1 Smallbone, Wheaton, & Hourigan (2003) Australia 43 35 Institution Moderate Yes
73.2 Smallbone & Milne (2000)
Table 1 continues.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 54
Table 1 continued.
PT
N N Degree Received?
73.3 Smallbone & McCabe (2003)
74.1 Smallbone & Wortley (2000) Australia 96 104 Institution Low Unknown
RI
74.2 Smallbone & Wortley (2004)
75 McKillop, Smallbone, Wortley, & Andjij (2012) Australia 55 49 Institution Low Unknown
SC
76 Hanson, Harris, Scott, & Helmus (2007) Canada 207 255 Community Moderate Mixed
77 Levenson (2004) United States 30 192 Institution High Mixed
NU
78 Hills (2002) Canada 43 50 Combined Moderate No
a
Study 48, Intrafamilial offenders had received treatment, while extrafamilial offenders had not received treatment.
MA
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 55
Table 2
Comparison between Intrafamilial and Extrafamilial Offenders Against Children on Demographic and Childhood History Variables
PT
Fixed-Effect Random-Effects Q I2 N (k)
RI
d [95% CI] d [95% CI]
Demographic Variables
SC
Young age .214 [.172, .255] .205 [.130, .280] 139.29*** 59.8% 10,019 (57)
Age at first sexual offense .685 [.581, .789] .657 [.356, .958] 67.49*** 86.7% 1,782 (10)
Age at first sexual offense outlier removed .507 [.393, .621] .517 [.363, .670] 12.66 36.8% 1,408 (9)
NU
Racial minority .211 [.096, .327] .179 [.032, .326] 26.94 29.5% 4,775 (20)
Homosexual or bisexual orientation 1.065 [.659, 1.470] 1.065 [.659, 1.470] 3.56 0.0% 625 (5)
MA
Unemployed .187 [.052, .322] .200 [-.061, .462] 31.96** 62.4% 1,686 (13)
Low income .182 [-.011, .375] .182 [-.011, .375] 7.36 0.0% 470 (9)
Employed in manual labour -.195 [-.564, .174] -.158 [-.641, .324] 4.76 37.0% 184 (4)
ED
Low education attainment -.128 [-.183, -.073] -.129 [-.213, -.044] 58.45** 41.8% 5,970 (35)
IQ test .149 [.055, .243] .117 [-.067, .302] 54.64*** 70.7% 2,430 (17)
Childhood History Variables
Childhood sexual abuse
Childhood physical abuse PT
-.106
-.078
[-.189, -.023]
[-.202, .046]
-.102
-.078
[-.201, -.002]
[-.202, .046]
22.77
8.34
16.5%
0.0%
4,578 (20)
1,635 (11)
CE
Family abuse -.312 [-.469, -.154] -.311 [-.471, -.150] 7.15 2.1% 912 (8)
Family neglect -.248 [-.487, -.008] -.172 [-.524, .181] 5.64 46.8% 275 (4)
Family problems outside of abuse .082 [-.046, .210] .140 [-.136, .417] 25.05*** 72.0% 1,230 (8)
AC
Poor childhood attachment with parents -.243 [-.451, -.036] -.243 [-.451, -.036] 1.59 0.0% 361 (5)
Poor childhood attachment with mother -.242 [-.480, -.003] -.242 [-.480, -.003] 1.69 0.0% 365 (5)
Poor childhood attachment with father -.181 [-.432, .071] -.182 [-.441, .077] 4.16 3.8% 357 (5)
Childhood internalizing behavior .007 [-.188, .203] .007 [-.188, .203] 1.94 0.0% 564 (3)
Note. A positive d indicates that extrafamilial offenders had more characteristics that were problematic (e.g., unemployment) or statistically rare
(e.g., childhood sexual abuse) than intrafamilial offenders. Bolded values indicate that the group differences were statistically significant, p < .05.
*p < .05. **p < .01. ***p < .001. Q refers to Cochran’s Q statistic, which is a statistical significance test for variability in the effect sizes across
studies. I2 is a measure of effect size for variability across studies, above what was expected by chance. I2 values of 25% are considered low,
50% moderate, and 75% high variability.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 56
Table 3
Comparison between Intrafamilial and Extrafamilial Offenders Against Children on Sexual Deviancy Variables
Fixed-Effect Random-Effects Q I2 N (k)
PT
Sexual Deviancy Variables d [95% CI] d [95% CI]
Prior sexual offensesa .427 [.362, .492] .471 [.355, .588] 90.42*** 61.3% 5,678 (36)
RI
Any paraphilias .293 [.218, .368] .351 [.227, .475] 64.28*** 54.9% 3.829 (30)
Pedohebephilia .408 [.332, .483] .396 [.248, .545] 70.25*** 67.3% 3,564 (24)
SC
Pedohebephilia outlier removed .323 [.243, .403] .337 [.223, .450] 34.53* 36.3% 3,179 (23)
Pedohebephilia assessed via PPG .328 [.238, .417] .328 [.238, .417] 11.24 0.0% 2,227 (15)
NU
Pedophilia .417 [.334, .500] .394 [.191, .596] 65.47*** 78.6% 2,700 (15)
Hebephilia .229 [.036, .422] .229 [.036, .422] 2.04 0.0% 458 (4)
Sexual interest in boys .470 [.232, .708] .372 [-.443, 1.187] 21.04*** 90.5% 301 (3)
MA
Sexual interest in girls .099 [-.076, .274] .118 [-.168, .404] 9.07 55.9% 543 (5)
Other paraphilias .627 [.493, .761] .605 [.216, .994] 59.21*** 86.5% 1,298 (9)
Other paraphilias outlier removed .488 [.348, .627] .448 [.262, .633] 10.68 34.4% 1,179 (8)
ED
Sexual self-regulation problems .134 [.035, .234] .170 [.008, .331] 23.12* 52.4% 1, 295 (12)
High sexual preoccupation .134 [.032, .236] .148 [-.022, .318] 20.55* 56.2% 1,752 (10)
PT
Short-term mating strategies -.029 [-.258, .201] .178 [-.348, .705] 6.27* 68.1% 395 (3)
Problems in the sexual sphere .070 [-.036, .177] .138 [-.063, .338] 44.33*** 68.4% 2,188 (15)
Low sexual knowledge .166 [-.031, .362] .196 [-.095, .488] 5.29 43.3% 430 (4)
CE
Sexual entitlement .107 [-.044, .257] .107 [-.044, .257] 1.47 0.0% 759 (3)
Other attitudes towards sex -.083 [-.278, .112] -.249 [-.716, .218] 16.27** 75.4% 477 (5)
AC
Other attitudes towards sex outlier removed -.001 [-.201, .198] -.001 [-.201, .198] 1.94 0.0% 451 (4)
Emotional congruence with children .393 [.282, .503] .465 [.277, .652] 14.12* 57.5% 1,307 (7)
Denial of sexual offense .048 [-.092, .187] .134 [-.191, .459] 30.99*** 77.4% 1,666 (8)
Denial of sexual offense outlier removed .272 [.103, .440] .246 [.017, .475] 9.36 35.6% 1,155 (7)
Minimization of sexual offense .194 [.094, .293] .345 [.039, .652] 36.00*** 80.6 2,440 (8)
Minimization outlier removed .123 [.021, .225] .123 [.021, .225] 2.79 0.0 2,321 (7)
Note. A positive d indicates that extrafamilial offenders had more characteristics that were inherently problematic (e.g., pedophilia) than
intrafamilial offenders. Bolded values indicate that the group differences were statistically significant, p < .05. *p < .05. **p < .01. ***p < .001. Q
refers to Cochran‘s Q statistic, which is a statistical significance test for variability in the effect sizes across studies. I2 is a measure of effect size
for variability across studies, above what was expected by chance. I2 values of 25% are considered low, 50% moderate, and 75% high variability.
a
Egger‘s intercept = 1.214 (p = .044). Trim and Fill method adjusted for 9 values and resulted in dfixed = 0.371, 95% CI = [0.308, 0.434], Trim and
Fill drandom = 0.370, 95% CI = [0.246, 0.493], Q = 134.36.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 57
Table 4
Comparison between Intrafamilial and Extrafamilial Offenders Against Children on Markers of Antisocial tendencies
Fixed-Effect Random-Effects Q I2 N (k)
PT
Markers of Antisocial Tendencies d [95% CI] d [95% CI]
Static-99 scores 1.353 [1.293, 1.413] 1.236 [1.041, 1.431] 70.86*** 83.1% 6,125 (13)
Any prior offenses .203 [.154, .252] .193 [.106, .281] 85.74*** 55.7% 9,097 (39)
RI
Prior violent offenses .161 [.077, .246] .151 [.016, .286] 30.09** 53.5% 3,393 (15)
SC
Callousness -.037 [-.119, .045] -.127 [-.446, .191] 119.23*** 90.8% 2,739 (12)
Callousness outlier removed .037 [-.047, .120] .066 [-.083, .216] 20.01* 50.0% 2,621 (11)
Lack of cooperation with supervision .141 [-.0004, .283] .196 [-.035, .427] 8.07 50.4% 1,012 (5)
NU
Self-regulation problems .102 [.016, .187] .102 [.010, .195] 20.81 8.7% 2,939 (20)
Impulsivity .082 [.001, .163] .082 [.001, .163] 17.37 0.0% 3,229 (19)
Poor problem solving .067 [-.074, .207] -.023 [-.266, .220] 10.53 52.5% 881 (6)
MA
Hostility .064 [-.004, .132] .076 [-.027, .179] 36.29* 42.1% 4,343 (22)
Hostility towards women .236 [.094, .378] .356 [-.159, .870] 46.53*** 89.2% 969 (6)
Low victim empathy 1.052 [.749, 1.355] 1.005 [.596, 1.414] 3.38 40.9% 192 (3)
ED
offense-supportive attitudes .178 [.096, .261] .239 [.008, .470] 82.42*** 81.8% 2,760 (16)
offense-supportive attitudes outlier removed .117 [.033, .201] .106 [-.014, .227] 19.82 29.4% 2,649 (15)
PT
Any substance misuse -.013 [-.075, .049] -.013 [-.075, .049] 24.48 0.0% 5,163 (26)
Substance use during offense .062 [-.086, .211] .057 [-.110, .224] 10.60 15.1% 1,607 (10)
Drug misuse .020 [-.063, .104] .011 [-.087, .110] 10.26 12.3% 3,146 (10)
CE
Alcohol misuse -.097 [-.169, -.025] -.097 [-.169, -.025] 13.72 0.0% 3,836 (16)
Employment problems .159 [-.004, .321] .149 [-.110, .408] 7.74 48.3% 755 (5)
AC
Negative peer group .173 [.089, .257] .360 [.053, .666] 62.55*** 90.4% 2,633 (7)
Negative peer group outliers removed .052 [-.038, .142] .052 [-.038, .142] 1.82 0.0% 2325 (5)
Antisocial .166 [.109, .223] .119 [.017, .222] 59.81*** 58.2% 5,752 (26)
PCL-R scores .223 [.151, .296] .216 [.089, .344] 24.80** 55.6% 3,713 (12)
PCL-R factor 1 (interpersonal/affective) .096 [-.001, .193] .020 [-.229, .269] 13.56** 70.5% 1,724 (5)
PCL-R factor 2 (behavioral) .255 [.155, .355] .255 [.155, .355] 3.07 0.0% 1,630 (5)
Childhood maladjustment -.050 [-.201, .101] -.037 [-.242, .168] 7.81 36.0% 969 (6)
Juvenile delinquency .087 [-.065, .239] .076 [-.285, .438] 25.07*** 76.1% 1,259 (7)
Note. A positive d indicates that extrafamilial offenders had more characteristics that were inherently problematic (e.g., antisocial tendencies) than intrafamilial
offenders. Bolded values indicate that the group differences were statistically significant, p < .05. *p < .05. **p < .01. ***p < .001. Q refers to Cochran’s Q
statistic, which is a statistical significance test for variability in the effect sizes across studies. I2 is a measure of effect size for variability across, above what was
expected by chance. I2 values of 25% are considered low, 50% moderate, and 75% high variability.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 58
Table 5
Comparison between Intrafamilial and Extrafamilial Offenders Against Children on Psychological Variables
I2
PT
Random-Effects Q N (k)
Variables d [95% CI] d [95% CI]
Interpersonal
RI
Any problems in the social sphere .121 [.058, .184] .224 [.076, .372] 102.08*** 74.5% 4,725 (27)
Social deficits .004 [-.080, .089] .004 [-.080, .089] 8.86 0.0% 2,629 (13)
SC
Loneliness .158 [.015, .302] .331 [-.023, .684] 17.80** 77.5% 910 (5)
Loneliness outlier removed .098 [-.049, .245] .132 [-.100, .364] 5.71 47.4% 726 (4)
NU
Poor social skills .176 [-.045, .398] .254 [-.196, .704] 17.39** 71.2% 332 (6)
Poor social skills outlier removed .081 [-.147, .310] .073 [-.243, .389] 6.66 40.0% 306 (5)
Detached romantic relationship style -.007 [-.090, .076] .031 [-.156, .219] 53.29*** 70.0% 3,342 (17)
MA
Detached outlier removed .030 [-.054, .114] .030 [-.054, .114] 14.54 0.0% 3,279 (16)
Intimacy problems .113 [-.059, .285] .347 [-.190, .883] 19.39*** 79.4% 738 (5)
Intimacy problems outlier removed .046 [-.129, .221] .046 [-.129, .221] 1.40 0.0% 712 (4)
ED
Never married .889 [.786, .991] 1.004 [.788, 1.219] 85.64*** 70.8% 4,067 (26)
Single .595 [.513, .678] .576 [.403, .748] 86.25*** 72.2% 4,517 (25)
Never cohabitated with a romantic partner 1.158 [.802, 1.514] 1.158 [.802, 1.514] 2.54 0.0% 538 (4)
Psychopathology
Anxiety .011 PT [-.068, .090] .096 [-.044, .235] 35.66** 52.3% 3,454 (18)
CE
Social anxiety .220 [-.140, .579] .358 [-.299, 1.016] 5.86 65.9% 133 (3)
Depression .005 [-.070, .080] .043 [-.066, .152] 29.86 33.0% 4,027 (21)
Repression -.246 [-.488, -.005] -.234 [-.520, .052] 2.74 27.1% 288 (3)
AC
General mental health issues -.025 [-.103, .053] .040 [-.072, .153] 31.10 29.3% 4,181 (23)
Personality disorder .109 [.017, .201] .109 [.017, .201] 4.82 0.0% 3,178 (12)
Poor coping skills .019 [-.084, .123] .172 [-.201, .545] 36.47*** 83.5% 2,244 (7)
Poor coping skills outliers removed -.004 [-.113, .104] .059 [-.132, .249] 5.49 27.2% 2029 (5)
Poor self-esteem -.044 [-.152, .064] -.040 [-.237, .156] 9.94 49.7% 1,330 (6)
Underassertiveness -.086 [-.188, .015] -.027 [-.300, .246] 19.73** 69.6% 1,905 (7)
Severe mental disorder .071 [-.009, .150] .071 [-.009, .150] 12.75 0.0% 3,366 (14)
Social desirability -.025 [-.143, .092] -.025 [-.219, .169] 39.15** 59.1% 1,436 (17)
Impression management -.024 [-.121, .074] -.002 [-.159, .155] 6.40 37.5% 2,019 (5)
Note. A positive d indicates that extrafamilial offenders had more characteristics that were inherently problematic (e.g., social deficits) than intrafamilial
offenders. Bolded values indicate that the group differences were statistically significant, p < .05. *p < .05. **p < .01. ***p < .001. Q refers to Cochran’s Q
statistic, which is a statistical significance test for variability in the effect sizes across studies. I2 is a measure of effect size for variability across studies, above
what was expected by chance. I2 values of 25% are considered low, 50% moderate, and 75% high variability.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 59
Table 6
Publication Bias and Study Year
Publication Bias Study Year
Variables Fixed effect
Q between p value published unpublished Z p value
Demographic Variables
PT
Young age 5.09 .024 0.26 0.17 -2.86 .004
Unemployed 0.001 .975 2.08 .038
Low education attainment 0.83 .362 -0.33 .740
RI
IQ test 0.05 .823 -2.02 .044
Family problems outside of abuse 0.08 .777 -0.58 .562
SC
Sexual Deviancy Variables
Prior sexual offenses 6.01 .014 0.52 0.35 0.81 .416
Any paraphilias 2.90 .088 -2.58 .010
Pedohebephilia (outlier removed) 0.17 .680 -0.76 .445
NU
Pedophilia 3.86 .050 0.84 .402
Sexual self-regulation problems 5.55 .018 0.35 0.07 -1.69 .092
High sexual preoccupation 5.65 .017 0.35 0.06 -1.78 .076
MA
Short-term mating strategies 1.28 .201
Problems in the sexual sphere 0.73 .393 -1.01 .311
Emotional congruence with children 0.13 .718 -2.10 .036
Markers of Antisocial Tendencies
D
Interpersonal Variables
Any problems in the social sphere 1.85 .174 -2.91 .004
Never married 4.63 .031 1.02 0.79 -3.49 <.001
Single 2.58 .108 -2.09 .036
Psychopathology variables
Anxiety 1.02 .312 -1.45 .147
Underassertiveness 3.39 .066 0.56 .573
Social desirability 0.62 .431 2.08 .038
Note. Bolded values reached p < .05. Moderator analyses comparing published to unpublished
study was conducted for each variable with statistically significant variability as indexed by the
Q statistics (outliers removed) and sufficient number of studies (at least 2 studies in each
category). Moderator analyses examining the influence of study year on effect sizes required at
least three studies.
ACCEPTED MANUSCRIPT
PUZZLE OF INTRAFAMILIAL SEXUAL ABUSE 60
Highlights
PT
Surprisingly few studies have examined family functioning variables
RI
SC
NU
MA
D
P TE
CE
AC