Content uploaded by Anthony F Bogaert
Author content
All content in this area was uploaded by Anthony F Bogaert on Apr 06, 2015
Content may be subject to copyright.
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 5, 2000
Fraternal Birth Order and Sexual Orientation
in Pedophiles
Ray Blanchard, Ph.D.,
1,2,4
Howard E. Barbaree, Ph.D.,
1,2
Anthony F. Bogaert, Ph.D.,
3
Robert Dickey, M.D., F.R.C.P.(C),
1,2
Philip Klassen, M.D., F.R.C.P.(C),
1,2
Michael E. Kuban, M.Sc.,
1
and Kenneth J. Zucker, Ph.D.
1,2
Whetherhomosexualpedophileshavemoreolderbrothers(ahigherfraternalbirth
order) than do heterosexual pedophiles was investigated. Subjects were 260 sex
offenders (against children age 14 or younger) and 260 matched volunteer con-
trols. The subject’s relative attraction to male and female children was assessed
by phallometric testing in one analysis, and by his offense history in another. Both
methods showed that fraternal birth order correlates with homosexuality in pe-
dophiles, just as it does in men attracted to physically mature partners. Results
suggest that fraternal birth order (or the underlying variable it represents) may
prove the first identified universal factor in homosexual development. Results also
argue against a previous explanation of the high prevalence of homosexuality in
pedophiles (25% in this study), namely, that the factors that determine sexual
preference in pedophiles are different from those that determine sexual preference
in men attracted to adults. An alternative explanation in terms of canalization of
development is suggested.
KEYWORDS: birth order;canalization;homosexuality;pedophilia;phallometry; sexualorientation;
sibling sex ratio.
INTRODUCTION
Studies of American, British, Canadian, and Dutch subjects have repeatedly
shownthat birthorder correlateswithsexualorientation inmen (Blanchard,1997);
1
Centre for Addiction and Mental Health, Toronto, Ontario, Canada.
2
Department of Psychiatry, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada.
3
Health Studies Program, Brock University, St. Catharines, Ontario, Canada.
4
To whom correspondence should be addressed at CAMH—Clarke Division, 250 College Street,
Toronto, Ontario M5T 1R8, Canada; e-mail: Ray
Blanchard@camh.net.
463
0004-0002/00/1000-0463$18.00/0
C
°
2000 Plenum Publishing Corporation
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
464 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
homosexual men have a higher mean birth order (i.e., more older siblings) than
comparable heterosexuals. It has further been shown that homosexual men have
a higher birth order than heterosexual men primarily because they have a greater
number of older brothers. They do not differwith regard to older sisters, oncetheir
number of older brothers has been taken into account (Blanchard and Bogaert,
1996a,b; Blanchard et al., 1998; Jones and Blanchard, 1998). This means that the
probability a man will be homosexual increases only in proportion to his number
of older brothers; older sisters neither increase nor decrease the probability of
homosexuality in later-born males. This phenomenon has therefore been termed
the fraternal birth order effect.
Virtually all this research has been carried out on men who are sexually
attractedtophysically mature partners.Onlytwostudies haveinvestigatedwhether
sexualorientationalsocorrelateswithfraternalbirthorderinpedophiles.Onestudy
foundthathomosexualpedophilesdohaveahigherfraternalbirthorder(moreolder
brothers) than heterosexual pedophiles (Bogaert et al., 1997); the other failed to
confirm this difference for men with sexual offenses against prepubescent boys
or girls, but did confirm it for men with offenses against pubescent boys or girls
(Blanchard and Bogaert, 1998).
The relation of fraternal birth order to sexualorientation among pedophilesis
relevant to theories of pedophilia and of homosexuality. The best epidemiological
evidence indicates that only 2–4% of men attracted to adults prefer men (ACSF
Investigators, 1992; Billy et al., 1993; Fay et al., 1989; Johnson et al., 1992);
in contrast, around 25–40% of men attracted to children prefer boys (Blanchard
et al.,1999; Gebhard etal., 1965; Mohret al., 1964).Thus, the rateof homosexual
attraction is6–20 timeshigher amongpedophiles. Oneexplanation ofthis discrep-
ancy is thatthe factors that determine sexual preference in pedophiles are different
from those that determine sexual preference in men attracted to adults (Freund
et al., 1984). A confirmed finding that fraternal birth order correlates with sexual
preference in both groups would indicate that this explanation is incorrect or, at
best, incomplete. Other hypotheses would have to be explored—for example, the
possibility that the same factors increase the probability of homosexuality in men
attracted to adults and men attracted to children, but they have a greater impact on
the latter.
The correlation of fraternal birth order and sexual orientation in pedophiles is
equally relevant to theories of homosexual development. No common feature has
yet been identified in the family demographics or developmental histories of all
typesof homosexualmen. The best established predictorof adult homosexuality—
cross-genderbehaviorinchildhood (Baileyand Zucker,1995)—does not appearto
characterize homosexual pedophiles (Freund and Blanchard, 1987; Freund et al.,
1982), although there is some evidence this group may be less masculine in boy-
hood without being more feminine (Freund and Blanchard, 1987). A finding that
high mean fraternal birth orders characterize men attracted to boys as well as those
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 465
attracted to other men—homosexual groups who differ as widely as possible in
theirowncharacteristics andinthe characteristics oftheir desired partners—would
suggest that fraternal birth order (or the underlying variable it reflects) may be the
first universal factor in homosexual development to be identified.
The inconsistency of the prior findings concerning the correlation of frater-
nal birth order and sexual orientation in pedophiles may relate to methodologic
problems in the two relevant studies. The first of these was a retrospective study
of sex offenders, which included only those subjects whose clinical charts hap-
pened to contain birth order data (Bogaertet al., 1997). Itis therefore possible that
some selection bias affected its results. The second was a reanalysis of archived
data from a classic, large-scale study of sexual offenders (Blanchard and Bogaert,
1998). The recoverable information regarding the subjects’ offense histories was
minimal, and it is possible that the sexual preferences of the pedophiles in that
study were not accurately classified from the available information.
Our study was therefore undertaken to settle the empirical question. This is
the first study in which data were collected with the express purpose of examining
the relation of fraternal birth order to sexual orientation in a consecutive series of
pedophiles.Itwasalsothefirsttoexaminedirectlytherelationbetweenpedophiles’
fraternalbirthordersandtheirpenileresponsestolaboratorystimulidepictingmale
and female adults and children.
METHOD
Subjects
There were two sources of pedophilic patients: Male outpatients referred to
the Kurt Freund Phallometric Laboratory of the Centre for Addiction and Mental
Health—Clarke Division (Toronto, Ontario) for psychophysiological assessment
of their erotic preferences, and incarcerated male sex offenders undergoing as-
sessment or treatment at the Sexual Behaviour Clinic of Warkworth Institution
(Campbellford, Ontario). A patient was selected for the present study if he met
one or more of the following inclusion criteria: (a) He had one or more charges,
convictions, or credible accusations of sexual interaction with boys or girls age 14
or under; (b) he had one or more self-reported instances of sexual interaction with
boys or girls age 14 or under; or (c) he verbally acknowledged that boys or girls
age 14 or under are more sexually interesting to him than men or women aged 17
or over. In the foregoing criteria, sexual interaction included noncontact offenses
(primarily exposing) as well as offenses involving physical touching.
Anotherwise eligiblepatient waseliminatedfrom thestudy ifhe meteither of
the following exclusion criteria: (a) He had anyincest offenses,defined as offenses
against biological children, stepchildren, or children living in the same household
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
466 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
toward whom he acted as a father; or (b) he had any offenses against males or
females age 15 or over. Two further exclusion criteria disqualified any subject
from the study, patient or control: (a) The subject was a twin, or (b) the subject
was not sure that he knew of all children born to his biological mother.
Of the 260 pedophilic subjects selected according to the foregoing criteria,
245 came from the Toronto site and 15 from the Campbellford site. All subjects
were examined from 1995 through 1998. Two of the 245 Toronto subjects had
been assessed at the Kurt Freund Phallometric Laboratory 10 and 16 years before,
respectively, and data from their earlier assessments had been included in the study
by Bogaert et al. (1997).
There were 14 patients who complained of erotic attraction to children (the
third inclusion criterion) but stated that they had never acted on their feelings.
The number and ages of the children involved in the offenses of the remainder are
inexact quantities, partly because these occasionally had to be estimated (if, for
example, a man exposed himself to a group of schoolchildren), and partly be-
cause some patients undoubtedly concealed offenses for which they were never
apprehended. The 246 patients with known offenses involved a total number of
913 children, for a mean of 3.71 children per patient (SD = 10.50). Children
<6 years old accounted for 12% of the total, children 6–11 years old accounted
for 69%, and children 12–14 years old accounted for 19%.
Control subjects were drawn from a pool of 663 men that included subjects
who had participated in an earlier study (Blanchard and Bogaert, 1996b) and
additional subjects who were recruited and examined in an identical manner to
augment that pool. These subjects were paid $10 (Canadian) for completing a
self-administered, anonymous questionnaire concerning their family background
and other personal information. They were recruited through clubs for promoting
business contacts, particular charities, or social events for specific communities
and similar organizations; and through advertisements, posted on two university
campuses, for heterosexual research subjects. Subjects recruited at club meetings
ororganizedcommunityeventswere solicited with the understandingthatpayment
wouldbemadetodesignatedcharitiesontheirbehalf;thoserecruitedasindividuals
were paid for their participation directly. All subjects indicated, in response to a
sexual-orientation item on the questionnaire, that they were heterosexual. It is
reasonable to assume, on purely statistical grounds, that virtually 100% of them
were gynephilic (i.e., most interested sexually in physically mature females), and
they are so labeled in this article. An elaborate matching procedure, which is
described later, was used to select 260 men from this pool. Of those selected, 162
had been in the earlier study and 98 had not.
Thematchingprocedureensuredthatthemeanageofthecontrols,33.97years
(SD = 12.83), was very close to that of the pedophilic patients,34.73 years (SD =
14.20). It was not feasible to match subjects on education, however,and the groups
did differ in this regard. Education was coded on an 8-point ordinal scale. The
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 467
medianpedophilicpatienthadsomehighschooleducation,butwithoutgraduation;
the median control subject had either completed or was currently enrolled in a
bachelor’sprogram. The educationaldifference between patients and controls was
significant [Mann–Whitney U(N = 520) = 8541.00, p <.001]. This statistical
test was two-tailed, as were all others reported in this article unless otherwise
specified. The groups did not differ significantly in ethnic origin: 87% of the
patients and 93% of the control subjects described themselves as White.
Materials and Procedure
Historical and Demographic Data
Information on a pedophilicpatient’s history of sexual offenses came primar-
ily from objective documents on his chart; for example, reports from probation
and parole officers. These offenseswere recorded on a standardized protocol form,
as were any additional offenses that he might admit. This form was also used to
record the patient’s subjectiveself-report regardingthe age and sexof persons who
most interest him sexually. For those procedures in which patients were selected
or classified on the basis of their sexual offenses, this was done on the basis of
all available information regarding their history of sexual offenses, not just on the
basis of their latest offense or some otherwise determined index offense. In such
procedures, patients with no sexual offenses were selected or classified according
to their self-report instead.
Information concerning personal demographics (e.g., a man’s age, educa-
tional level, ethnic origin) and family demographics (e.g., the ages of the man’s
parents at the time of his birth, the sequence of boys and girls in his sibship) was
collected from both patients and controls by means of a self-administered ques-
tionnaire. Only children born to the subject’s biological mother were recorded;
that is, maternal half-siblings were recorded but paternal half-siblings were not.
There were both theoretical and methodologic reasons for this: (1) maternal half-
siblings—like full siblings—develop in the same uterus as the subject, whereas
paternal half-siblings do not; (2) maternal half-siblings are more likely than pater-
nal half-siblings to be reared in the same home as the subject; and (3) a subject is
more likely to know of all children carried by his mother than he is to know of all
children sired by his father. In this study, maternal half-siblings were counted the
same as full siblings.
Phallometric Measures
For the pedophilic patients at the Toronto site, phallometric testing was used
to assess the subject’s attraction to male versus female prepubescent children,
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
468 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
male versus female pubescent children, and male versus female adults. In the
phallometric test procedure, penile blood volume, the dependent measure of erotic
arousal, is monitored during the presentation of potentially erotic test stimuli. The
testusedinthis study is amodificationof one described indetailelsewhere(Freund
and Blanchard, 1989). The stimuli are audiotaped narratives presented through
headphones and accompanied by slides. There are seven categories of narratives,
which describe sexual interactions with prepubescent girls, pubescent girls, adult
women, prepubescent boys, pubescent boys, and adult men, and also solitary,
nonsexual activities (“neutral” stimuli). All narratives are written in the second
person and present tense (e.g., “You are babysitting your neighbors’ little girl for
the evening”) and are approximately 100 words long. The narratives describing
heterosexual interactions are recorded with a woman’s voice, and those describing
homosexual interactions, with a man’s. Neutral stimuli are recorded with both.
Each test trial consists of one narrative, accompanied by photographic slides
showingthefrontview,rearview,andgenitalregionofanudemodelcorresponding
in age and sex to the topic of the narrative. Neutral narratives are accompanied by
slides of landscapes. The full test consists of four blocks of seven trials, with each
block including one trial of each type in fixed pseudorandom order. The length of
eachtrial is54 sec;however,the intervalbetween trials variesbecause penileblood
volume must return to its baseline (flaccid) value before a new trial is started. The
time required to complete a test is usually about 1 hr.
Penile blood volume change is sampled four times per second throughout a
trial.Thesubject’sresponse is quantifiedintwoways:astheextremumofthecurve
of blood volume change (i.e., the greatest departure from initial value occurring
during the 54 sec of the trial), and as the area under the curve. To identify subjects
whose penile blood volume changes might reflect random physiological variation
ratherthan eroticarousal,themeanofthethreehighestpositiveextremumscores—
aquantity called the Output Index(Freund,1967)—is calculated. The phallometric
data of subjects who fail to meet the criterion output index of 1.0 ml are excluded.
Each subject’s 28 extremum scores are then converted into standard scores,
based only on his own extremum data, and the same operation is carried out
on his area scores. Next, for each subject, the standardized extremum and area
scores are combined to yield a separate composite score for each of the 28 trials,
using the formula, (z
E
i
+ z
A
i
)/2, where z
E
i
is the standardized extremum score
for the ith trial, and z
A
i
is the standardized area score for the ith trial. These
operationsarecarriedout for the followingreasons:(a) In phallometricwork,some
transformationofrawscoresis generally requiredin combining datafrom different
subjects, because the between-subjects variability in absolute magnitude of blood
volumechanges can otherwise obscure evenquite reliable statistical effects.There
are numerous sources of such variability; for example, the subject’s age, his state
of health, the size of his penis, and the amount of time since his last ejaculation
from masturbation or interpersonal sexual activity. Empirical research has shown
the z-score transformation to be satisfactory (Harris et al., 1992; Langevin, 1985).
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 469
(b) The (highly correlated) area and extremum z-scores are averaged to obtain a
composite that reflects both the speed and amplitude of response and lessens the
impact of anomalous responses; that is, large change from initial value but small
area or vice versa (Freund et al., 1983).
In the last stage of basic processing, the data are reduced to seven final scores
for each subject by averaging his four composite scores in each of the seven
stimulus categories. These seven category scores are taken as measures of the
subject’s relative erotic interest in adult women, pubescent girls, prepubescent
girls, and so on.
For the present study, the previously mentioned category scores were used to
computethreespecialmeasuresof sexualorientation.A Pedophilic Sex-Preference
Index was calculated as the subject’s category score for prepubescent boys minus
his category score for prepubescent girls, and a Hebephilic Sex-Preference Index
was similarly calculated as the subject’s category score for pubescent boys minus
his category score for pubescent girls. Finally, a Teleiophilic Sex-Preference Index
(Greek t
´
eleios, “full-grown”) was calculated as the subject’s category score for
adult men minus his category score for adult women.
It should be noted that many pedophilic subjects can and do control their re-
actions during phallometric testing in order to produce a more “normal” response
profile (e.g., Freund and Blanchard, 1989). This is primarily accomplished by sup-
pressing responses to pubescent and prepubescent stimulus categories. There was
no reason to expect, however, that such attempts at test manipulation might pro-
duce spurious positivecorrelations between the subject’s number of older brothers
and his scores on the Pedophilic or Hebephilic Sex-Preference indices. On the
contrary, the most likely result of such efforts would be to diminish the apparent
size of true positive correlations.
Consent
All of the patients and the supplemental volunteer subjects signed consent
forms giving permission for their data to be used for research purposes. Control
subjects for the earlier study (Blanchard and Bogaert, 1996b) had instead been
given an information sheet that statedthat the purpose of the study was to examine
the relation between people’s family backgrounds and their later sexual orienta-
tion as adults. That alternative procedure had been approved by the University of
Toronto Human Subjects Review Committee, which also examined and approved
the information sheet.
RESULTS
The previously described version of the phallometric test was administered
to 214 of the Toronto patients. The data of 18 were discarded because of technical
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
470 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
Table I. Linear Regressions of the Three Sex-Preference Indices on the Subject’s Number of Older
Brothers, Older Sisters, Younger Brothers, and Younger Sisters
Sex-preference indices (criteria)
Pedophilic Hebephilic Teleiophilic
Sibling-type (predictors) β p β p β p
Older brothers .26 .001 .15 .07 .20 .01
Older sisters −.06 .47 −.04 .64 −.11 .17
Younger brothers .01 .93 −.02 .78 .03 .72
Younger sisters .09 .25 .12 .12 .03 .70
problems (e.g., the patient was too obese to get the volumetric apparatus properly
seated) or gross uncooperativeness from the patient (e.g., he continued to avoid
looking at the visual stimuli, despite repeated instructions to attend to them). The
data of another 6 were discarded because the patient responded insufficiently to
meet the Output Index criterion.
The phallometric data of the remaining 190 patients were examined in a
series of linear regression analyses. A separate analysis was carried out for each
of the three sex-preference indices, which served as the criterion variables. All
analyses used the same four predictor variables—namely, the patient’s number of
older brothers, older sisters, younger brothers, and younger sisters. The predictors
were entered directly into the regression equation. The results (specifically, the
standardized β coefficients and their associated significance levels) are shown in
Table I.
A positive β coefficient in Table I means that a greater number of siblings (of
a particular type) was associated with a higher (more homosexual) score on the
corresponding sex-preference index. Number of older brothers correlated with the
Pedophilic Sex-Preference Index and, to a lesser degree but still significantly, with
the Teleiophilic Sex-Preference Index. The correlation between number of older
brothers and the Hebephilic Sex-Preference Index was in the same direction, but
this correlation did not reach the .05 level of statistical significance. None of the
other three types of siblings correlated significantly with any of the sex-preference
indices. These results indicate that the more older brothers a pedophile has, the
greater his sexual interest in boys compared with girls.
Two additional linear regression analyses were conducted, both using the
Pedophilic Sex-Preference Index as the criterion variable. The first of these added
the age of the patient’s mother at the time of the patient’s birth as a fifth predictor
variable; the second added the age of the patient’s father instead. The patients
in the first analysis were those 175 who knew their mother’s age, and those in
the second were the 165 who knew their father’s age. The results did not change
any of the conclusions. The number of older brothers, and only the number of
older brothers, was significant in both analyses. Neither maternal nor paternal
age correlated with the relative preference for male vs female children (β = .01,
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 471
p = .90, and β = .08, p = .33, respectively). These results demonstrate that the
observed fraternal birth order effect was not really an artifact of parental age.
Thelinear regressionanalysesof phallometricdata were a sensitivemeans for
showing that sexual orientation correlates with number of older brothers among
pedophiles as it does among teleiophiles. These analyses did not, however, show
whether homosexual pedophiles (conceived as a discrete group) have more older
brothers than men in the general population, or where heterosexual and bisexual
pedophiles stand in this regard. Another analysis was therefore carried out, using
the already-mentioned gynephilic control group.
In this analysis, the full group of 260 Toronto and Campbellford pedophiles
were divided into three groups according to their offense histories or self-reported
eroticpreferences:152heterosexualpedophiles(menwithoffensesorself-reported
attractionsinvolvinggirlsonly),43bisexualpedophiles(boysandgirls),and65ho-
mosexual pedophiles (boys only). These were matched with an equal number of
gynephilic controls, bringing the total number of subjects in this analysis to 520.
The gynephilic controls were matched to the pedophilic patients on two vari-
ables. The first of these was the subject’s year of birth. This variable was used
because demographic trends during the years when a sample of randomly selected
subjects were born can have a significant effect on that sample’s expected birth
order (Hare and Price, 1969, 1974; Price and Hare, 1969). The second variable
was the subject’s total number of siblings other than older brothers—that is, older
sisters plus younger brothers plus younger sisters. This variable was used as a
means of controlling for family size while allowing the subject’s number of older
brothers to varybetween groups. The decision to add these three classes of siblings
together was justified by the findings of the linear regression analyses that none of
them correlated with sexual orientation, once number of older brothers was taken
into account.
The 260 controls were randomly selected from the pool of 663 gynephilic
volunteers so as to match the joint distribution of pedophilic patients by year of
birthand number of siblingsotherthanolderbrothers.Thiswaseffectedas follows:
The combined group of pedophilic patients was dividedinto quartiles according to
year of birth. These quartiles were cross-tabulated with the subject’s total number
of siblings other than older brothers. The same procedure was carried out on the
pool of gynephilic volunteers. From each cell in the volunteers’ table, a sample of
controls, equal in size to the number of patients in the corresponding cell in the
patients’ table, was selected. This is most easily understood by example: There
were 13 patients and 29 gynephilic volunteers who were born between 1952 and
1962, and who had a total of 3 siblings other than older brothers. A computer
program therefore randomly selected 13 of the 29 volunteers for the control group.
When the number of available volunteers in a cell was smaller than the number of
patients, adjacent cells in the volunteers’ table were combined.
The control group selected by this procedure was closely similar to the com-
bined group of pedophiles on the two relevantvariables.The mean year of birth for
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
472 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
Fig. 1. The mean number of siblings of each type reported by each group. “Gyne
Controls,” gynephilic controls; “Hetero Pedo,” heterosexual pedophiles; “Bisex Pedo,”
bisexual pedophiles; “Homo Pedo,” homosexual pedophiles.
the controls was 1960.73 (SD = 13.84) and that for the pedophiles was 1961.57
(SD = 14.26). The controls’ mean number of siblings other than older brothers
was 2.29 (SD = 1.98), and that of the patients was 2.28 (SD = 1.99).
Figure 1 shows the mean numbers of older brothers, older sisters, younger
brothers, and younger sisters for each of the four groups. One would expect, both
from the results of the regression analyses and from the fact that these variables
were used in selecting the matched controls, that there would be no differences
betweengroupsinmeannumbers of older sisters,youngerbrothers,oryounger sis-
ters. This was confirmed by one-way ANOVAs (all Fs < 0.40). In contrast, a one-
way ANOVA for number of older brothers, the “unconstrained” variable, revealed
the presence of significant between-groups differences (F(3, 516) = 4.20, p =
.006). A set of Dunnett tests, which compared each of the three pedophilic groups
with the gynephilic control group, showed that the homosexual pedophiles had
moreolderbrothers than thecontrols (p = .002),butthe heterosexualand bisexual
pedophiles did not (p = .29, and p = .99, respectively).These results confirm the
finding of the regression analyses that fraternal birth order correlates with sexual
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 473
orientation in pedophiles; they further suggest that fraternal birth order does not
correlate with pedophilia per se.
A final series of analyses investigated whether the sibships of any of the
groups contained an unusual excess of males or females. The variable, sibling sex
ratio, is commonly reported as the ratio of brothers to sisters collectively reported
by a given group of subjects (who are usually referred to as the probands in this
type of research). In White populations, the ratio of male live births to female
live births is close to 106 males per 100 females (Chahnazarian, 1988; James,
1987). In the computation of inferential statistics, this value is more conveniently
expressed as the proportion of males rather than the ratio of males to females,
that is, .515 (106/206). For reasons pertaining to the existence of slight between-
couplesvariabilityintheprobabilityofproducingmaleoffspring,anadjustedvalue
of .518 has been suggested for use when a sample of probands consists entirely of
males (James, 1998).
In this study, the gynephilic controls had 380 brothers and 368 sisters, for a
sibling sex ratio of 103 (and a proportion of .508); the heterosexual pedophiles
had 243 brothers and 228 sisters, for a sibling sex ratio of 107 (.516); the bisexual
pedophiles had 57 brothers and 58 sisters, for a sibling sex ratio of 98 (.496);
and the homosexual pedophiles had 123 brothers and 94 sisters, for a sibling sex
ratio of 131 (.567). These data were compared with the adjusted population value
(.518) using the z approximation to the binomial test. The sibling sex ratios of
the first three groups showed no evidence of a greater than expected proportion of
brothers (all p values >.30, one-tailed). The sibling sex ratio of the homosexual
pedophiles, however, did approach statistical significance (p = .08, one-tailed).
This result is probably not redundant with the finding of a high fraternal birth
order for this group, because the sex ratio was high for siblings born after the
proband (123) and for those born before the proband (137).
DISCUSSION
Implications for Sexual Orientation
The results show that pedophiles with more older brothers have more sexual
interest in boys relative to their sexual interest in girls. These results confirm
the prior finding that sexual orientation correlates with fraternal birth order in
pedophiles (Bogaert et al., 1997) and thus establish that this correlation extends
beyond men who prefer physically mature partners (teleiophiles).
The present study did not address the specific mechanism linking older broth-
ers and homosexuality. The authors have previously speculated that the proximate
cause is maternal antibodies to Y-linked minor histocompatibility antigens (H-Y
antigens), which are raised in increasing concentrations by each succeeding male
fetus, and which have increasingly stronger effects on sexual differentiation in the
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
474 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
brainin eachsucceedingmalefetus(BlanchardandBogaert,1996b;Blanchard and
Klassen, 1997). This hypothesis is consistent with a variety of evidence, includ-
ing the apparent irrelevance of older sisters to the sexual orientation of later-born
males, the absence of any correlation between birth order and sexual orientation
in females, the probable involvement of H-Y antigen in the development of sex-
typical traits, and the detrimental effects of immunization of female mice to H-Y
antigenon the reproductiveperformance of subsequent male offspring(Blanchard,
1997; Blanchard and Klassen, 1997).
There are, of course, other possible explanations of the fraternal birth order
effect besides the maternal immune hypothesis (e.g., Bem, 1996; Sulloway, 1996,
pp. 433–434, 488). A variety of these has been reviewed elsewhere (Blanchard,
1997). The most popular rival hypothesis is the notion that sexual interaction with
older males increases a boy’s probability of developing a homosexual orientation,
and that a boy’s chances of engaging in such interactions increase in proportion
to his number of older brothers (e.g., Jones and Blanchard, 1998). Although this
hypothesis may seem intuitively plausible, there are little empirical data to recom-
mend it (see discussion in Purcell et al., in press).
Implications for Pedophilia
The proportion of pedophiles in this study who were exclusively or primarily
interested in boys, as assessed from their offense histories, was 25%. This result
is consistent with previous studies that suggest the prevalence of homosexuality is
about 10 times higher in pedophiles than in teleiophiles (Blanchard et al., 1999;
Gebhard et al., 1965; Mohr et al., 1964). As previously noted, one explanation of
this discrepancy is that the factors that determine sexual preference in pedophiles
are different from those that determine sexual preference in teleiophiles (Freund
etal., 1984).The finding that fraternal birth order correlateswith sexualpreference
in both groups shows that this explanation is incorrect or, at best, incomplete.
An alternative explanation for the high prevalence of homosexuality in pe-
dophiles is the following: Pedophilia and homosexuality tend to occur in the same
men because these individuals are generally less resistant to factors that divert
psychosexual development from the species-typical outcome of sexual interest in
receptive, physically mature females. This hypothesis is consistent with evidence
that other sexually variant behaviors, for example, exhibitionism, are also com-
mon in pedophiles (Paitich et al., 1977; Raymond et al., 1999; Rooth, 1973). This
hypothesis proposes that the clustering of sexual variations results from the ab-
sence of a single protective factor, rather than the presence of multiple pathogenic
factors, or a single pathogenic factor with multiple effects. The missing protective
factor could be analogous to, or an aspect of, the biological phenomenon known
as canalization—that is, the tendency for feedback loops to return a developing
system to its usual pathway, when that system has been diverted to a minor extent
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 475
from it (e.g., Waddington, 1962). As used by biologists, this term includes mecha-
nisms that stabilize development against disruptive environmental influences and
thosethatbufferdevelopmentagainstgenetic-mutationalloss of function (Wilkins,
1997). Of particular relevance to the present discussion is the finding that there
are individual differences in efficiency of canalization, with some persons being
more prone to multiple developmental disruptions than others (Bogle et al., 1994;
Cronk and Reed, 1981; Prader et al., 1963; Rose et al., 1987; Shapiro, 1975).
The hypothesis that the failure of some protective (or corrective) mechanism
might account for the emergence of multiple paraphilias has previously been ad-
vancedinothercontexts.Forexample,theclusteringoffetishism,transvestism,and
autogynephilia (sexual arousal in association with thoughts of having a woman’s
body) has been attributed to “the failure of some developmental process that, in
normal males, keeps heterosexual learning ‘on track’” (Blanchard, 1991, p. 247).
Inthe present case, the correlation of pedophiliaand mentalretardation (Blanchard
et al., 1999) suggests that the hypothesized failure of canalization may sometimes
be part of a larger picture of neurological dysfunction.
The canalization-failure hypothesis—or any similar hypothesis that focuses
on the susceptibility of the individual—can explain why the prevalence of ho-
mosexuality is higher in pedophiles than in teleiophiles, even if the factors that
influence sexual orientation in the two groups are identical in nature and in objec-
tive magnitude. This hypothesis changes the original question from, “Why is the
prevalence of homosexuality so high in pedophiles?” to “Why do the same indi-
viduals tend to become both homosexual and pedophilic?” The greatest drawback
of the hypothesis is its silence regarding the difference in cross-gender behavior
between boyswho will later be homosexual pedophiles and those who will later be
attracted to adult men—a difference that was handily explained by the hypothesis
that homosexuality has different etiologies in pedophiles and teleiophiles (Freund
et al., 1984). This issue may prove a fatal flaw in the canalization-failure hypoth-
esis, leading to its eventual rejection, or merely a temporary lack that is filled by
subsequent theoretical elaboration.
Implications for Gender Identity Disorder
The topic of cross-gender behavior is also raised by one of this study’s sec-
ondary findings. Severalstudies of homosexual men, including some ofthe largest
studies of homosexuality ever conducted, have found that homosexual men have
higher than expected sibling sex ratios. This research has been reviewed in detail
elsewhere (Blanchard, 1997). In the authors’ experience, homosexual groups who
have a sibling sex ratio that is significantly higher than the expected value of 106
always have an excess of younger brothers as well as an excess of older brothers,
althoughtheexcessofyoungerbrothers is considerablysmaller(Blanchard, 1997).
This suggests that a high sibling sex ratio is not merely an alternative view of a
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
476 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
high fraternal birth order, but perhaps a separate phenomenon requiring its own
explanation (Blanchard, 1997).
An early review of the evidence (Blanchard et al., 1995) suggested that high
sibling sex ratios are specific to extremely feminine homosexual male groups
(e.g., transsexuals). Several studies in the succeeding years seemed to confirm that
hypothesis (Blanchard, 1997). The latest evidence, however,casts doubt on it. One
recent study found a high sibling sex ratio in a sample of predominantly African-
Americanhomosexualmen,whowerenotrecruitedwithregardto their gender role
behavior, and who may or may not have been notably feminine (Bogaert, 1998).
That result is consistent with the present finding of a sibling sex ratio of 131 for
homosexual pedophiles (who, as already explained, are not feminine). That value
was not statistically significant given the sample size (N = 217 siblings), but it
is quite high in absolute terms. Moreover, the sibling sex ratio was high for the
subjects’younger siblings and for their older siblings. The present study,therefore,
also suggests that the hypothesis of Blanchard et al. (1995) was incorrect.
Implications for Societal Attitudes
A few closing comments are necessary to preclude any misunderstanding
or misuse of this study. First, the statistical association of homosexuality and
pedophilia concerns developmental events in utero or in early childhood. Ordinary
(teleiophilic) homosexual men are no more likely to molest boys than ordinary
(teleiophilic) heterosexual men are to molest girls. Second, the causes of homo-
sexuality are irrelevant to whether it should be considered a psychopathology.
That question has already been decided in the negative, on the grounds that homo-
sexuality does not inherently cause distress to the individual or any disability in
functioning as a productive member of society (Friedman, 1988; Spitzer, 1981).
ACKNOWLEDGMENTS
Study supported in part by Standard Research Grant 410-95-0003 from the
Social Sciences and Humanities Research Council of Canada to Ray Blanchard.
The authors thank Heike Boedeker, James M. Cantor, Martin L. Lalumi`ere,
Timothy Perper, Michael C. Seto, and Cathy Spegg for their various forms of
assistance.
REFERENCES
ACSF Investigators. (1992). AIDS and sexual behaviour in France. Nature 360: 407–409.
Bailey, J. M., and Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A
conceptual analysis and quantitative review. Dev. Psychol. 31: 43–55.
Bem, D. J. (1996). Exotic becomes erotic: A developmental theory of sexual orientation. Psych. Rev.
103: 320–335.
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
Sexual Orientation in Pedophiles 477
Billy, J. O. G., Tanfer, K., Grady, W. R., and Klepinger, D. H. (1993). The sexual behavior of men in
the United States. Fam. Plann. Perspect. 25: 52–60.
Blanchard, R. (1991). Clinical observations and systematic studies of autogynephilia. J. Sex Marital
Ther. 17: 235–251.
Blanchard, R. (1997). Birth order and sibling sex ratio in homosexual versus heterosexual males and
females. Ann. Rev. Sex Res. 8: 27–67.
Blanchard, R., and Bogaert, A. F. (1996a). Biodemographic comparisons of homosexual and hetero-
sexual men in the Kinsey interview data. Arch. Sex. Behav. 25: 551–579.
Blanchard, R., and Bogaert, A. F. (1996b). Homosexuality in men and number of older brothers. Am.
J. Psychiat. 153: 27–31.
Blanchard, R., and Bogaert, A. F. (1998). Birth order in homosexual versus heterosexual sex offenders
against children, pubescents, and adults. Arch. Sex. Behav. 27: 595–603.
Blanchard, R., and Klassen, P. (1997). H-Y antigen and homosexuality in men. J. Theor. Biol. 185:
373–378.
Blanchard, R., Watson, M. S., Choy, A., Dickey, R., Klassen, P., Kuban, M., and Ferren, D. J. (1999).
Pedophiles: Mental retardation, maternal age, and sexual orientation. Arch. Sex. Behav. 28: 111–
127.
Blanchard, R., Zucker, K. J., Bradley, S. J., and Hume, C. S. (1995). Birth order and sibling sex ratio
in homosexual male adolescents and probably prehomosexual feminine boys. Dev. Psychol. 31:
22–30.
Blanchard, R., Zucker, K. J., Siegelman, M., Dickey, R., and Klassen, P. (1998). The relation of birth
order to sexual orientation in men and women. J. Biosoc. Sci. 30: 511–519.
Bogle, A. C., Reed, T., and Rose, R. J. (1994). Replication of asymmetry of a–b ridge count and
behavioral discordance in monozygotic twins. Behav. Genet. 24: 65–72.
Bogaert, A. F. (1998).Birth order and sibling sex ratio in homosexual andheterosexualnon-white men.
Arch. Sex. Behav. 27: 467–473.
Bogaert, A. F., Bezeau, S., Kuban, M., and Blanchard, R. (1997). Pedophilia, sexual orientation, and
birth order. J. Abnorm. Psychol. 106: 331–335.
Chahnazarian, A. (1988). Determinants of the sex ratio at birth: Review of recent literature. Soc. Biol.
35: 214–235.
Cronk, C. E.,and Reed, R.B. (1981). Canalizationof growth in Down syndromechildren three months
to six years. Hum. Biol. 53: 383–398.
Fay, R. E., Turner, C. F., Klassen, A. D., and Gagnon, J. H. (1989). Prevalence and patterns of same-
gender sexual contact among men. Science 243: 338–348.
Freund, K. (1967). Diagnosing homo- or heterosexuality and erotic age-preference by means of a
psychophysiological test. Behav. Res. Ther. 5: 209–228.
Freund, K., and Blanchard, R. (1987). Feminine gender identity and physical aggressiveness in het-
erosexual and homosexual pedophiles. J. Sex Mar. Ther. 13: 25–34.
Freund, K., and Blanchard, R. (1989). Phallometric diagnosis of pedophilia. J. Consult. Clin. Psychol.
57: 100–105.
Freund, K., Heasman, G., Racansky, I. G., and Glancy, G. (1984). Pedophilia and heterosexuality vs.
homosexuality. J. Sex Mar. Ther. 10: 193–200.
Freund, K., Scher,H., Chan, S., and Ben-Aron, M. (1982). Experimental analysisof pedophilia. Behav.
Res. Ther. 20: 105–112.
Freund, K., Scher, H., and Hucker, S. (1983). The courtship disorders. Arch. Sex. Behav. 12: 369–379.
Friedman, R. C. (1988). Male Homosexuality: A Contemporary Psychoanalytic Perspective, Yale
University Press, New Haven, CT.
Gebhard, P. H., Gagnon, J. H., Pomeroy, W. B., and Christenson, C. V. (1965). Sex Offenders: An
Analysis of Types, Harper & Row, New York.
Hare, E. H., and Price, J. S. (1969). Birth order and family size: Bias caused by changes in birth rate.
Br. J. Psychiat. 115: 647–657.
Hare, E. H., and Price, J. S. (1974). Birth order and birth rate bias: Findings in a representative sample
of the adult population of Great Britain. J. Biosoc. Sci. 6: 139–150.
Harris, G. T., Rice, M. E., Quinsey, V. L., Chaplin, T. C., and Earls, C. (1992). Maximizing the
discriminant validity of phallometric assessment data. Psych. Assess. 4: 502–511.
James, W. H. (1987). The human sex ratio. Part 1: A review of the literature. Hum. Biol. 59: 721–
752.
P1: FMN
Archives of Sexual Behavior [asb] PL112-54 July 29, 2000 11:20 Style file version Nov. 19th, 1999
478 Blanchard, Barbaree, Bogaert, Dickey, Klassen, Kuban, and Zucker
James, W. H. (1998).The sex ratios of thesibs of childrenwith psychiatric diagnoses.J. Child Psychol.
Psychiat. 39: 595.
Johnson, A. M., Wadsworth, J., Wellings, K., Bradshaw, S., and Field, J. (1992). Sexual lifestyles and
HIV risk. Nature 360: 410–412.
Jones, M. B., and Blanchard, R. (1998). Birth order and male homosexuality: An extension of Slater’s
Index. Hum. Biol. 70: 775–787.
Langevin, R. (1985). Introduction. In Langevin, R. (ed.), Erotic Preference, Gender Identity, and
Aggression in Men: New Research Studies, Lawrence Erlbaum Associates, Hillsdale, NJ, pp. 1–
13.
Mohr,J.W., Turner,R.E.,and Jerry, M. B. (1964). Pedophilia and Exhibitionism, UniversityofToronto
Press, Toronto, Ontario, Canada.
Paitich, D., Langevin, R., Freeman, R., Mann, K., and Handy, L. (1977). The Clarke SHQ: A clinical
sex history questionnaire for males. Arch. Sex. Behav. 6: 421–436.
Prader, A., Tanner, J. M., and von Harnack, G. A. (1963). Catch-up growth following illness or starva-
tion: An example of developmental canalization in man. J. Pediatr. 62: 646–659.
Price, J. S., and Hare, E. H. (1969). Birth order studies: Some sources of bias. Br. J. Psychiat. 115:
633–646.
Purcell, D. W., Blanchard, R., and Zucker, K. J. (in press). Birth order in a contemporary sample of
gay men. Arch. Sex. Behav.
Raymond, N. C., Coleman, E., Ohlerking, F., Christenson, G. A., and Miner, M. (1999). Psychiatric
comorbidity in pedophilic sex offenders. Am. J. Psychiat. 156: 786–788.
Rooth, G. (1973). Exhibitionism, sexual violence and paedophilia. Br. J. Psychiat. 122: 705–710.
Rose, R. J., Reed, T., andBogle, A.(1987). Asymmetryof a–b ridge count and behavioral discordance
of monozygotic twins. Behav. Genet. 17: 125–140.
Shapiro, B. L. (1975). Amplified developmental instability in Down’s syndrome. Ann. Hum. Genet.
38: 429–437.
Spitzer,R.L. (1981). The diagnostic status of homosexualityinDSM-III:Areformulationofthe issues.
Am. J. Psychiat. 138: 210–215.
Sulloway, F. J. (1996). Born to Rebel: Birth Order, Family Dynamics, and Creative Lives, Pantheon
Books, New York.
Waddington, C. H. (1962). New Patterns in Genetics and Development, Columbia University Press,
New York.
Wilkins, A. S. (1997). Canalization: A molecular genetic perspective. BioEssays 19: 257–262.
