Content uploaded by Brian B Boutwell
Author content
All content in this area was uploaded by Brian B Boutwell on Oct 07, 2017
Content may be subject to copyright.
ORIGINAL PAPER
Sexual Orientation and Involvement in Nonviolent and Violent
Delinquent Behaviors: Findings From the National Longitudinal
Study of Adolescent to Adult Health
Kevin M. Beaver
1,2
•Eric J. Connolly
3
•Joseph A. Schwartz
4
•Brian B. Boutwell
5,6
•
J. C. Barnes
7
•Joseph L. Nedelec
7
Received: 30 January 2014 / Revised: 27 January 2016 / Accepted: 12 February 2016 / Published online: 7 April 2016
Springer Science+Business Media New York 2016
Abstract This study examined the association between sexual
orientation and nonviolent and violent delinquency across the life
course. We analyzed self-reported nonviolent and violent delin-
quency in a sample of heterosexual males (N=5220–7023) and
females (N=5984–7875), bisexuals (N=34–73), gay males (N=
145–189), and lesbians (N=115–150) from the National Lon-
gitudinal Study of Adolescent to Adult Health (Add Health). The
analyses revealed, in general, that bisexuals were the most delin-
quent of the sexual orientation categories for both males and
females. Additional analyses revealed that heterosexual males re-
ported significantly higher levels of both violent and nonviolent
delinquency than gay males, whereas lesbians reported more invol-
vement in nonviolent delinquency and, to a lesser extent, violent
delinquency relative to heterosexual females. Analyses also revea-
led that lesbians reported significantly more delinquent behavior,
particularly for nonviolent delinquency, than gay males. Future re-
search should explore the mechanisms that account for these
observed patterns and how they can be used to more fully under-
stand the etiology of delinquency.
Keywords Add Health Antisocial behavior
Delinquency Sexual orientation Violence
Introduction
One of the more intriguing findings to emerge from crimi-
nological research is the consistent association between various
dimensions of sexuality and involvement in an assortment of
antisocial behaviors (Bell, O’Neal, Feng, & Schoenrock, 1999;
Harden, Mendle, Hill, Turkheimer, & Emery, 2007; Lussier,
Proulx, & LeBlanc, 2005; Nedelec & Beaver, 2012;Widom,
1977; Zuckerman, Bone, Neary, Mangelsdorff, & Brustman,
1972; Zuckerman, Tushup, & Finner, 1976). In general, ado-
lescents and adults who report a greater number of sexual
partners (Boutwell, Barnes, Deaton, & Beaver, 2013;Nedelec
& Beaver, 2012;Widom,1977; Zuckerman et al., 1972,1976),
who have a n earlier age of sexual debut (Armour & Haynie, 2007;
Ramrakha et al., 2007), and who engage in riskier sexual behav-
iors (Biglan et al., 1990; Metzler, Noell, Biglan, Ary, & Smolk-
owski, 1994) are comparatively more likely to have a criminal
record, to self-report greater involvement in crime and delinqu-
ency, and to be rated as more antisocial.
The nexus between sexual behaviors and criminal involve-
ment is robust, having been detected across a wide range of hete-
rogeneous studies. To illustrate, Ellis and Walsh (2000)exam-
ined 51 studies that tested for an association between the number
of sexual partners and criminal behavior and 50 of them reported
a significant association. Moreover,theyalsoreviewed31stud-
ies that had examined the association between age of first sexual
experience and antisocial behavior. All 31 studies included in
their review found a statistically significant association, wherein
&Kevin M. Beaver
kbeaver@fsu.edu
1
College of Criminology and Criminal Justice, Florida State
University, Tallahassee, FL 32306-1127, USA
2
Center for Social and Humanities Research, King Abdulaziz
University, Jeddah, Saudi Arabia
3
Department of Criminal Justice, Pennsylvania State University,
Abington, Abington, PA 19001, USA
4
School of Criminology and Criminal Justice, University of
Nebraska at Omaha, Omaha, NE 68588-0561, USA
5
Criminology & Criminal Justice, School of Social Work, Saint
Louis University, St. Louis, MO 63103, USA
6
Department of Epidemiology, Saint Louis University, St. Louis,
MO 63104, USA
7
School of Criminal Justice, University of Cincinnati,
Cincinnati, OH 45221-0389, USA
123
Arch Sex Behav (2016) 45:1759–1769
DOI 10.1007/s10508-016-0717-3
an earlier onset of sexual behavior was associated with greater
involvement in antisocial behavior. Based on the existing find-
ings, the link between sexual involvement and antisocial behav-
iors (such as crime, delinquency, drug use, and certain deviant be-
haviors) appears to be firmly established.
Although the available empirical evidence clearly reveals that
overall patterns of sexual behavior are significantly related to a
wide range of criminal and antisocial behaviors, there remain a
number of issues that may be connected to the sexual behavior–
crime association that have yet to be fully explored. Perhaps, the
area of research that is most lacking in this line of inquiry centers
on the (potential) association between sexual orientation and
criminal or delinquent involvement. Ellis, Hoffman, and Burke
(1990) analyzed the association between sexual orientation and
criminal and violent behaviors in a sample of male and female
university students. The results revealed that heterosexual males
were more criminal and violent than gay males on 7 of the 26
behaviors examined (significant correlations ranged between
-.19 and -.41), but that bisexual males were more criminal and
violent than heterosexual males (as indicated by significant mean
differences). Additionally, Ellis et al. reported that lesbians were
more criminal and violent than heterosexual females on 5 of the
26 behaviors examined (significant correlations ranged between
.16 and .26), but not as criminal or as violent as heterosexual
males.
A more recent study also reported significant associations be-
tween sexual orientation and convictions for delinquency. In this
analysis, Garnette, Irvine, Reyes, and Wilber, (2011)reported
that 13 % of incarcerated adolescent males were gay and about
23 % of incarcerated females were lesbians. These estimates are
greater than what is typically found in the general population,
thereby suggesting that homosexuals are disproportionately more
involved in acts of crime and delinquency or that they are more
likely to be apprehended and processed through the criminal
justice system.
These within-sex sexual orientation differences in delinqu-
ency conform to a broader set of findings of within-sex sexual
orientation differences in sex-typed behavior (Sandfort, 2005).
For instance, gay men are typified as being more feminine and
less masculine than heterosexual males, while the opposite is
true for lesbians when compared to heterosexual females (Lippa,
2008).
1
Thus, the findings reported previously showing differ-
ences in delinquency across sexual orientation groups are con-
sistent with this overall pattern. Even so, Koeppel (2015) gener-
ated results contradictory to those reported previously. Data were
drawn from 645 college students and these students were asked
questions about their criminal behaviors and their sexual orienta-
tion. The results of the multivariate analyses revealed no signifi-
cant association between sexual orientation and assault, theft, and
total number of crimes the respondent reported committing. In
addition, Koeppel examined the potential association between
sexual orientation and levels of self-control. Once again, the re-
sults revealed no significant differences among sexual orientation
groups in terms of their average levels of self-control (B=.001,
p[.05). However, one potential explanation for this disjunction
with prior research may be that Koeppel grouped together bisex-
ual, lesbian, and gay participants into one category labeled non-
heterosexuals. Such categorization leaves the potential differential
association between sexual orientation and criminal behavior
among heterosexual, gay, lesbian, and bisexual individuals
unaddressed.
The limited available research thus provides mixed evidence
regarding the association between sexual orientation and anti-
social behaviors. The studies that did detect a significant asso-
ciation were generated from convenience samples of non-rep-
resentative populations (e.g., college students and inmates) or
samples that are now outdated. This leaves open the possibility
that the findings might not generalize to other samples or more
contemporary samples. As a result, more research is needed to
address the potential nexus between sexual orientation and anti-
social behaviors. The current study addressed this gap in the lit-
erature by examining the link between sexual orientation and
nonviolent and violent criminal behaviors in adolescence and
young adulthood for both males and females using data from a
nationally representative sample of youth. Based on the limited
available research, we expected the cascade of antisocial behavior
to be arranged as follows: heterosexual males[lesbians[gay
males[heterosexual females, without making any predictions
with respect to bisexuals.
Method
Participants
Data for this study were drawn from the National Longitudinal
Study of Adolescent to Adult Health (Add Health) (Udry, 2003).
Add Health is a four-wave nationally representative sample of
American youth who were enrolled in middle or high school
during the 1994–1995 school year. Data collection began with
Wave 1 in-school surveys. These surveys were completed by all
students who were in attendance on a specified day at each of the
132 schools that were included in the study. In total, more than
90,000 students completed the Wave 1 in-school surveys. A
subsample of these youth was then selected to be re-interviewed
in their homes along with their primary caregivers. These sur-
veys, which are now referred to as the Wave 1 in-home surveys,
weredesignedtoaskmoredetailedquestionsandtocoverawider
range of topics, including some that were sensitive in nature.
Participants, for example, were asked about their family li fe, their
involvement in delinquent behaviors, and their use of drugs and
alcohol. Importantly, for each wave, participants provided their
answers to these sensitive questions on a laptop computer and not
1
We wishto thank the Editorfor pointing outthis relativelyparsimonious
explanation.
1760 Arch Sex Behav (2016) 45:1759–1769
123
to the interviewer present in their home. Overall, 20,745 ado-
lescents and more than 17,000 of their primary caregivers par-
ticipated in the Wave 1 in-home component to the study (Harris,
2009;Harrisetal.,2003).
About 1.5 years later, the second round of surveys was admin-
istered to 14,738 of the original Wave 1 in-home participants.
Since most of the participants were still adolescents, the survey
instruments remained relatively unchanged from the Wave 1 in-
home surveys. Once again, participants were asked a broad range
of questions, including those related to their sexual behaviors,
their family and peer relationships, and their involvement in risky
behaviors. In 2001 and 2002, the third wave of data was collected.
At this time, most of the participants were young adults and thus
the questionnaires were redesigned to include more age-appro-
priate questions. For example, the participants were asked about
their marital status, their educational and work history, and their
experience with childrearing. Overall, 15,197 participants were
included in the Wave 3 component of the study. The fourth wave
of data was collected in 2007–2008 and a total of 15,701 par-
ticipants participated. During this wave, most of the participants
were between the ages of 24 and 32 years and they were again
asked a broad array of questions, including those related to their
involvement in criminal behaviors, their highest level of edu-
cation, and their victimization experiences.
Measures
Sexual Orientation
We opted to measure sexual orientation in adulthood because
previous Add Health research has shown that items pertaining to
sexual orientation in adolescence likely yield unreliable and
invalid results (Savin-Williams & Joyner, 2013). To circumvent
this issue, we elected to focus our attention on sexual orientation
in adulthood, a time period when participants are more likely to
be comfortable with their sexuality and less likely to lie or mislead
others about their sexual preferences. During Wave 4 interviews,
participants were asked to self-report on their sexual orientation.
They were provided with the following responses: 100 % hetero-
sexual (straight); mostly heterosexual (straight), but somewhat at-
tracted to people of your own sex; bisexual (that is, attracted to men
and women equally); mostly homosexual (gay), but somewhat at-
tracted to people of the opposite sex; 100 % homosexual (gay); and
not sexually attracted to either males or females (asexual). Overall,
85.64 % of the sample indicated that they were 100 % heterosex-
ual, 9.72 % indicated that they were mostly heterosexual, 1.58 %
indicated that they were bisexual, 0.83 % indicated that they were
mostly homosexual, 1.33 % indicated that they were 100 % homo-
sexual, 0.45 % indicated that they were not attracted to either males
or females, and about 0.40 % did not respond.
Two different measures of sexual orientation were used in the
current analysis. First, the original categories outlined above
were used to examine the association between sexual orientation
and involvement in delinquency. Importantly, those partici-
pants who indicated that they were asexual were removed from
the analyses because there were too few of them to calculate
ANOVAs that provided stable estimates (some cells included
zeroes). Second, another sexuality measure was created by
pooling together the 100 % heterosexual and mostly hetero-
sexual categories, eliminating the bisexual and asexual cat-
egories, and pooling together the 100 % homosexual category
with the mostly homosexual category. The sexuality measure
was then transformed intoa dichotomous variable, whereinthe
two outcomes were heterosexual/mostly heterosexual (97.8 %)
and homosexual/mostly homosexual (2.2 %).
Delinquency
At each wave of data collection, participants were asked a series
of questions pertaining to their involvement in a variety of delin-
quent behaviors. As a result, and following previous research
(Haynie & South, 2005; Mears, Cochran, & Beaver, 2013), we
were able to create three delinquency measures at each wave: a
nonviolent delinquency scale, a violent delinquency scale, and a
total delinquency scale that combined the nonviolent and violent
delinquency scales into a composite measure. The nonviolent
delinquency scales included questions that asked the participant
to indicate how many times in the past 12 months they had
engaged in a series of behaviors, such as damaging property,
stealing a car, stealing money, and acting rowdy (responses were
coded so that 0 =never, 1 =one or two times, 2 =three or four
times, and 3 =five or more times). Responses to these items were
then summed together to create the Wave 1 (alpha =.80; absolute
range =0–27), the Wave 2 (alpha =.80; absolute range =0–27),
theWave3(alpha=.73; absolute range =0–20), and the Wave 4
(alpha =.69; absolute range =0–23) nonviolent delinquency
scales.
A similar procedure was used to create the violent delinqu-
ency scales. Again, at each wave, participants were asked how
many times in the past 12 months they had engaged in a series of
behaviors, such as getting into a group fight, getting into a phys-
ical fight, and injuring someone badly enough that they needed
medical care (responses for most of the items were coded so that
0=never, 1 =one or two times, 2 =three or four times, and
3=five or more times; for two items [whether the participants
shot or stabbed someone and whether they pulled a knife or gun
on someone], the item was coded such that 0 =no and 3 =at least
once).Responsestotheseitemswerethensummedtocreatethe
Wave 1 (alpha =.78; absolute range =0–18), Wave 2 (alpha =.79;
absolute range =0–18), Wave 3 (alpha =.64; absolute range =
0–14), and Wave 4 (alpha =.60; absolute range =0–13) violent
delinquency scales. Similar scales have been used in previous
research examining violent delinquency with the Add Health
data (Haynie & South, 2005; Mears et al., 2013).
Arch Sex Behav (2016) 45:1759–1769 1761
123
The items from the wave-specific nonviolent and violent del-
inquency scales were summed together to create the Wave 1
(alpha =.86; absolute range =0–45), Wave 2 (alpha =.85;
absolute range =0–45), Wave 3 (alpha =.76; absolute range =
0–27), and Wave 4 (alpha =.72; absolute range =0–24) total
delinquency scales. Higher values on these scales represent greater
involvement in nonviolent and violent delinquent behaviors dur-
ing the preceding 12 months.
In addition to these wave-specific delinquency scales, we also
developed a total nonviolent delinquency scale and a total violent
delinquency scale that captured involvement in delinquency
across all four waves of measurement. The total nonviolent delin-
quency scale was created by summing together all four of the
wave-specific nonviolent delinquency scales (alpha =.80) and
the total violent delinquency scale was created by summing
together all four of the wave-specific violent delinquency scales
(alpha =.84). These two scales provide total measures of delin-
quent involvement during adolescence and young adulthood.
Results
Delinquency and Sexual Orientation for Males (5
Groups)
We first compared mean differences on nonviolent, violent, and
total composite delinquency scales across sexual orientation for
males by estimating ANOVAmodels. These resultsare shown
in Table 1. For nonviolent delinquency, there were significant
differences across the five sexual orientation categories for all
four-wave-specific scales (Fvalues ranged between 4.54 and
10.46, p\.05). In general, bisexual males were the most delin-
quent, followed by mostly heterosexual and heterosexual males.
Males who identified as mostly gay and gay were, for the most
part, the least involved in nonviolent delinquency during all four
waves. A relatively similar pattern of results was detected for
violent delinquency at Waves 1 and 2 (F=11.25 and F=4.17,
respectively; p\.05). Once again, bisexual males were the most
violent at these two waves, while gay males tended to be the least
violent. There were no statistically significant differences across
the sexual orientation groups for Wave 3 or Wave 4 violent
delinquency (F=2.04 and F=2.17, respectively; p[.05),
which is likely a function of the reduced mean and variance
observed for all groups at these later waves (notice that all of the
means are below 1.00). The bottom panel of Table 1shows the
findings for the total composite delinquency scales. For all four of
these scales, there were average group differences that were
detected (Fvalues ranging between 5.42 and 7.56, p\.05). In
line with the previous findings, bisexual males were, on average,
the most involved in delinquency followed by mostly hetero-
sexual and heterosexual males. Mostly gay males and gay males
were, on average, the least involved in delinquency across these
four waves of data. Appendix 1includes post hoc paired com-
parisons for all of the sexual orientation groups (for males) for all
of the outcome measures, showing which groups differed sig-
nificantly from each other (Bonferroni corrected for multiple
comparisons).
Table 1 ANOVA tests for mean values on delinquency scales across sexual orientation for males
Heterosexual Mostly heterosexual Bisexual Mostly gay Gay Fvalue
Nonviolent delinquency
Wave 1 2.70 3.36 3.98 1.95 1.73 5.83*
Wave 2 2.08 2.81 3.11 1.30 1.45 4.54*
Wave 3 0.81 1.67 1.15 0.66 0.71 10.46*
Wave 4 0.38 0.84 0.68 0.67 0.37 9.29*
Violent delinquency
Wave 1 1.70 1.33 1.83 0.50 0.38 11.25*
Wave 2 1.14 1.36 1.57 0.37 0.43 4.17*
Wave 3 0.38 0.40 0.18 0.15 0.14 2.04
Wave 4 0.34 0.46 0.46 0.65 0.18 2.17
Total composite delinquency
Wave 1 4.39 4.61 5.81 2.45 2.09 7.56*
Wave 2 3.20 4.21 4.58 1.67 1.88 5.42*
Wave 3 1.18 2.08 1.32 0.81 0.82 7.17*
Wave 4 0.76 1.34 1.18 1.33 0.59 6.51*
Due to contractual requirements of the Add Health and concerns about deductive disclosure, we did not report the Ns for each cell in this table because
some of the cells had relatively small sample sizes that would lead to situations where we would be unable to ensure respondent anonymity on these
sensitive topics
* Indicates significant mean differences at the p\.05 level (equal variances not assumed) with Bonferroni correction applied (new plevel is\.004)
1762 Arch Sex Behav (2016) 45:1759–1769
123
Delinquency and Sexual Orientation for Females (5
Groups)
The second set of analyses pertained to females. Table 2shows
the results of these analyses: across all of the nonviolent, violent,
and total composite scales, there were statistically significant
average differences in delinquency across the sexual orientation
groups (Fvalues ranging between 6.88 and 55.60, p\.05). In
general, bisexual females, on average, self-reported the greatest
involvement in all forms of delinquency. Mostly heterosexual,
mostly lesbian, and lesbians scored, on average, the next highest
on the delinquency scales, though the exact ordering fluctuated
depending on the delinquency scale being analyzed. Hetero-
sexual females were, for the most part, the least involved in de-
linquency across each of the delinquency measures. Appendix 2
includes post hoc paired comparisons for all of the sexual ori-
entation groups (for females) for all of the outcome measures,
showing which groups differed significantly from each other
(Bonferroni corrected for multiple comparisons).
Delinquency and Sexual Orientation for Males (2
Groups)
We next examined the results wherein we excluded bisexual
males and females and only focus on participants who fell in
the outer ranges of the sexual orientation continuum. Specifi-
cally, we compared participants who indicated that they were
heterosexual or mostly heterosexual against those who indicated
that they were gay or mostly gay. The results for males are shown
in Table 3. It can be seen that across the 12 mean comparisons,
there were 8 statistically significant differences (tvalues ranging
between 2.77 and 14.42, p\.05). For all of these statistically
significant differences, heterosexual males exhibited the highest
levels of delinquency, including being more heavily involved in
both nonviolent and violent forms of delinquency. None of the
Wave 4 delinquency scales differed significantly between hetero-
sexual males and gay males.
Delinquency and Sexual Orientation for Females (2
Groups)
For females, Table 4shows a very different pattern of findings.
Specifically, 9 of the 12 mean comparisons were statistically
significant (tvalues ranging between -2.81 and -2.29, p\.05),
but this time lesbians were more involved in delinquent behaviors
than were heterosexual females. Moreover, the results revealed
the most pronounced differences in self-reports of nonviolent de-
linquency, with all four of the wave-specific nonviolent delin-
quency scales being significantly different between heterosexual
females and lesbians. Interestingly, only one of the violent
delinquency scales (Wave 1) was significantly different between
heterosexual females and lesbians (t=-2.34, p\.05). None of
the other violent delinquency scales were significantly different
Table 2 ANOVA tests for mean values on delinquency scales across sexual orientation for females
Heterosexual Mostly heterosexual Bisexual Mostly lesbian Lesbian Fvalue
Nonviolent delinquency
Wave 1 1.64 2.80 2.87 2.78 2.45 54.05*
Wave 2 1.28 2.40 2.34 2.30 2.00 44.99*
Wave 3 0.25 0.62 0.71 0.41 1.06 39.15*
Wave 4 0.12 0.43 0.63 0.45 0.35 55.60*
Violent delinquency
Wave 1 0.72 0.91 1.31 1.16 1.27 11.06*
Wave 2 0.45 0.64 0.85 0.64 0.63 6.88*
Wave 3 0.06 0.10 0.17 0.14 0.32 8.29*
Wave 4 0.10 0.18 0.21 0.05 0.40 8.14*
Total composite delinquency
Wave 1 2.36 3.71 4.18 3.95 3.72 43.34*
Wave 2 1.73 3.03 3.19 2.95 2.66 37.29*
Wave 3 0.31 0.72 0.89 0.55 1.39 38.22*
Wave 4 0.23 0.66 0.79 0.57 0.79 43.88*
Due to contractual requirements of the Add Health and concerns about deductive disclosure, we did not report the Ns for each cell in this table because
some of the cells had relatively small sample sizes that would lead to situations where we would be unable to ensure respondent anonymity on these
sensitive topics
* Indicates significant mean differences at the p\.05 level (equal variances not assumed) with Bonferroni correction applied (new plevel is\.004)
Arch Sex Behav (2016) 45:1759–1769 1763
123
between the two groups (tvalues ranging between -1.94 and
-1.00).
Delinquency and Gay Males vs. Lesbians
We next compared the means between gay males and lesbians.
Table 5shows the results of these analyses and, as can be seen, 5
of the 12 mean comparisons were significantly different. Speci-
fically, there were significant mean differences for the Wave 1
and Wave 2 nonviolent delinquency scales (t=-2.21 and t=
-2.22, respectively; p\05), the Wave 1 violent delinquency
scale (t=-3.72, p\.05),and the Waves 1 and 2 total composite
delinquency scales (t=-3.18 and t=-2.23, p\.05). Of par-
ticular salience is that in all of these instances, it was lesbians
who self-reported greater involvement in acts of delinquency.
Delinquency, Heterosexuals, Gay Males, and Lesbians
The last two sets of analyses used the total nonviolent delin-
quency scale and the total violent delinquency scale and exam-
ined the mean levels for heterosexual males, gay males, hete-
rosexual females, and lesbians. Figure1plots the means and
95 % confidence intervals for the total nonviolent delinquency
scale. An Ftest (F=104.86, p\.05) indicated that these group
means were significantly different from each other. Moreover,
the figure shows that heterosexual males scored the highest on
total nonviolent delinquency, followed closely by lesbians, then
gay males, and finally by heterosexual females.
A very similar pattern was detected when examining the total
violent delinquency scale. Figure 2plots the mean levels of vio-
lent delinquency and shows that, once again, these mean levels
differed significantly across the male/female sexuality categories
(F=178.82, p\.05). Heterosexual males had the highest mean
score, followed by lesbians, then heterosexual females, and finally
gay males.
Discussion
The results of the current study revealed three main findings. First,
there were significant mean differences in delinquent involve-
ment across the sexual orientation categories for both males and
females. For the most part, bisexual males and females scored
the highest on the examined delinquency scales (though some
exceptions did emerge). Additionally, and as Table 3reveals,
heterosexual males, in comparison to gay males, self-reported
Table 3 Mean differences in delinquency between heterosexual and gay males
Nonviolent delinquency Violent delinquency Total delinquency
Heterosexuals Gay males Heterosexuals Gay males Heterosexuals Gay males
Wave 1 2.72 1.80 1.69 0.42 4.40 2.20
SD 3.86 2.98 2.75 1.13 5.82 3.51
N6937 187 6948 189 6920 187
tvalue 4.17* 14.42* 8.27*
df 203.21 253.68 214.56
Wave 2 2.10 1.41 1.14 0.41 3.23 1.82
SD 3.39 2.27 2.36 1.08 5.00 2.80
N5242 145 5237 145 5220 145
tvalue 3.58* 7.63* 5.83*
df 162.27 184.28 170.72
Wave 3 0.84 0.70 0.38 0.14 1.21 0.82
SD 1.89 1.51 1.15 0.54 2.51 1.74
N5561 161 5622 163 5539 161
tvalue 1.22 5.30* 2.77*
df 174.74 206.87 179.82
Wave 4 0.40 0.46 0.35 0.33 0.78 0.83
SD 1.23 1.44 1.18 1.52 1.99 2.78
N7019 189 7023 189 7014 189
tvalue -0.57 0.12 -0.25
df 195.50 178.60 177.81
The gay male category was created by collapsing the homosexual and mostly homosexual categories and the heterosexual category was created by
collapsing the heterosexual and mostly heterosexual categories
* Indicates significant mean difference at the p\.05 level, two-tailed ttest (equal variances not assumed)
1764 Arch Sex Behav (2016) 45:1759–1769
123
significantly greater involvement in violent delinquency across
three of the four waves of data and nonviolent delinquency in two
of the four waves of data collection. The opposite pattern was
detected with females wherein lesbians, not heterosexual females,
self-reported significantly greater involvement in nonviolent delin-
quency at all four waves and greater involvement in violent delin-
quency in one of the four waves. The second key finding to emerge
from the analyses was that, when gay males and lesbians were com-
pared, lesbians reported significantly greater involvement in nonvi-
olent delinquency at two waves and violent delinquency at one wave;
none of the remaining differences were statistically significant. Third,
results from the analyses examining total nonviolent and total violent
delinquency across heterosexual males and females and gay males
and lesbians revealed that heterosexual males reported the highest
levels of nonviolent delinquency, followed closely by lesbians, then
gay males, and finally heterosexual females. For violent delinquency,
heterosexual males were again the most delinquent, followed by
lesbians, then heterosexual females, and finally gay males.
These findings have implications for research on crime and
delinquency in general, as well as male–female differences
in antisocial behavior. One of the—if not the—most consistent
findings in relation to criminal behavior is that males are sig-
nificantly more violent, aggressive, and antisocial than females
(Barnes, Jorgensen, Beaver, Boutwell, & Wright, 2015; Moffitt,
Caspi, Rutter, & Silva, 2001; Wilson & Herrnstein, 1985). This
finding has been detected across a diverse set of samples, across
different types of societies, and across broad time periods
(Cantor, 1982; Gottfredson & Hirschi, 1990;Moffittetal.,2001;
Wilson & Herrnstein, 1985). This study is unique, therefore,
because it reveals that the male–female difference may not be
universal and that sexual orientation plays an important role.
While the male–female difference observed in prior work is not
in dispute, what remains to be determined is the potential mech-
anism(s) that might be able to explain such differences in off-
ending (Bennett, Farrington, & Huesmann, 2005;Rowe,A.T.
Flannery, & D. J. Flannery, 1995). Unfortunately, the results of the
current study do not provide any insight into the specific mecha-
nisms that might account for why there are significant differences
in delinquent involvement across sexual orientation categories.
This drawback should be addressed in future research in order to
uncover the potential biological influences (e.g., prenatal hormone
exposure), cultural effects, social-psychological processes, or so-
cialization patterns that might be central to both sexual orientation
and delinquency. For example,it would be interesting to examine
Table 4 Mean differences in delinquency between heterosexual females and lesbians
Nonviolent delinquency Violent delinquency Total delinquency
Heterosexuals Lesbians Heterosexuals Lesbians Heterosexuals Lesbians
Wave 1 1.83 2.61 0.75 1.22 2.57 3.83
SD 2.82 3.60 1.66 2.44 3.87 5.43
N7862 148 7875 150 7851 148
tvalue -2.62* -2.34* -2.81*
df 150.41 151.62 149.83
Wave 2 1.46 2.15 0.48 0.64 1.93 2.80
SD 2.54 2.95 1.37 1.69 3.35 3.99
N5989 116 5999 115 5984 115
tvalue -2.49* -1.00 -2.31*
df 118.32 116.90 117.11
Wave 3 0.31 0.76 0.07 0.24 0.38 1.00*
SD 1.05 1.85 0.43 0.95 1.24 2.59
N6674 120 6685 122 6651 120
tvalue -2.68* -1.94 -2.63*
df 120.40 121.91 120.00
Wave 4 0.17 0.40 0.11 0.23 0.30 0.68
SD 0.77 1.27 0.64 0.94 1.13 1.84
N7919 150 7095 133 7094 133
tvalue -2.29* -1.49 -2.42*
df 151.05 134.31 133.88
The lesbian category was created by collapsing the homosexual and mostly homosexual categories and the heterosexual category was created by
collapsing the heterosexual and mostly heterosexual categories
* Indicates significant mean difference at the p\.05 level, two-tailed ttest (equal variances not assumed)
Arch Sex Behav (2016) 45:1759–1769 1765
123
0
2
4
6
Gay
Males
Heterosexual
Females
Heterosexual
Males
Lesbians
Violent Delinquency
Fig. 2 Mean scores on the total violent delinquency scale across sexual
orientation for males and females. Note 95 % confidence intervals shown
with error bars; Means significantly different at the p\.05 level, two-
tailed ttest (F=178.82)
0
2
4
6
Gay
Males
Heterosexual
Females
Heterosexual
Males
Lesbians
Nonviolent Delinquency
Fig. 1 Mean scores on the total nonviolent delinquency scale across
sexual orientation for males and females. Note 95 % confidence intervals
shown with error bars; Means significantly different at the p\.05 level,
two-tailed ttest (F=104.86)
Table 5 Mean differences in delinquency between gay males and lesbians
Nonviolent delinquency Violent delinquency Total delinquency
Gay
males
Lesbians Gay
males
Lesbians Gay
males
Lesbians
Wave 1 1.80 2.61 0.42 1.22 2.20 3.83
SD 2.98 3.60 1.13 2.44 3.51 5.43
N187 148 189 150 187 148
tvalue -2.21* -3.72* -3.18*
df 283.51 198.90 239.66
Wave 2 1.41 2.15 0.41 0.64 1.82 2.80
SD 2.27 2.95 1.08 1.69 2.80 4.00
N145 116 145 115 145 115
tvalue -2.22* -1.22 -2.23*
df 211.77 184.73 196.41
Wave 3 0.70 0.76 0.14 0.24 0.82 1.00
SD 1.51 1.85 0.54 0.95 1.74 2.59
N161 120 163 122 161 120
tvalue -0.30 -1.00 -0.66
df 225.75 178.93 196.07
Wave 4 0.46 0.40 0.33 0.23 0.83 0.68
SD 1.44 1.27 1.52 0.94 2.78 1.84
N189 150 174 133 174 133
tvalue 0.41 0.71 0.56
df 332.90 293.59 299.41
The gay and lesbian categories were created by collapsing the homosexual and mostly homosexual categories
* Indicates significant mean difference at the p\.05 level, two-tailed ttest (equal variances not assumed)
1766 Arch Sex Behav (2016) 45:1759–1769
123
whether key measures derived from the minority stress model
would account for part of this association (Meyer, 1995), which
is a particularly interesting possibility given that stress appears to
be related to criminal involvement (Agnew, 1992). This is just
one of several explanations that might be able toaccount for part
of the reason why there is a connection between sexual orienta-
tion and criminal behavior. Future research would benefit greatly
by empirically testing measures derived from different theories
and explanations to determine which factors might underlie the
sexual orientation–delinquency association.
It is also worth noting that our statistical models were able to
partially account for male–female differences in nonviolent and
violent delinquency. To our knowledge, very few studies have
shown that males and females do not differ significantly in terms of
their self-reported violence. Moreover, even fewer studies are
available that indicate females are more violent than males. Our
results showed, however, that lesbians were either significantly
more violent or nonviolent than gay males or that there were no
significant differences between the two groups. Further research is
needed to more fully address the factors that might be able to ac-
count for such differences. Nonetheless, the key point is that the
widely accepted sex gap in offending may be conditional on sexual
orientation.
Although our findings converged with those of some previ-
ous studies (Ellis et al., 1990), interpretation of the current results
needs to be viewed with caution due to at least two limitations.
First, the measures of nonviolent and violent delinquency were
based on self-reports, not official arrest data. This leaves open
the possibility that participants were not forthcoming about their
involvement in acts of delinquency. While this could be the case,
previous research has shown that self-reports of crime and delin-
quency tend to be reliable and valid ways of capturing variation
in antisocial behavior (Thornberry & Krohn, 2000). Pollock,
Menard, Elliott, and Huizinga (2015) recently reported a strong
agreement between self-reported arrests and officially recorded
arrests. When there was disagreement between the two reporting
sources, the data revealed that participants were just as likely to
report an arrest that did not show up in official records as they
were to lie in the opposite direction. In other words, there is no
reason to suspect that participants systematically under-report
their involvement in criminal activity in surveys. This means
that it would require a unique and nuanced hypothesis to explain
why we should observe the present results simply as a result of
unreliable measurement strategies. If one were to argue that our
use of self-reports has artificially imposed group differences,
then one would need to have a theoretical justification for why
lesbians would be more likely to over-report delinquency com-
pared to gay males (the same could be said for the other signif-
icant differences that emerged). While we are not ruling out this
possibility completely, it is important to note that a methodo-
logical artifact is highly unlikelytohaveresultedinconsistent
differences across groups.
A second limitation to keep in mind when interpreting the
results is that previous research has revealed the measurement of
sexual orientation in adolescence with the Add Health is prob-
lematic and may not be particularly valid (Savin-Williams &
Joyner, 2013). We were able to overcome this issue by measuring
sexual orientation from questions drawn in adulthood (at Wave
4), not adolescence. It is quite possible that measuring sexual
orientation in the Add Health at Wave 4 is more reliable because
sexual orientation is more likelytoremainstableonceonere-
aches adulthood. At the same time, participants were probably
more likely to be forthcoming about their sexual orientation in
adulthood than they would be during adolescence. Even so, it
would be important for future research to examine theconnection
between sexual orientation and criminal behaviors at numer-
ous time points throughout the life course. This type of research
design would reveal whether changes in sexual orientation cor-
respond to changes in criminal/delinquent involvement and
would help address whether this association is causal.
There has been a tremendous amount of scholarship focused
on studying and attempting to explain male–female differences
in antisocial behaviors. Unfortunately, most of this research has
overlooked the importance of simultaneously examining how
sexual orientation might fit into the equation. The findings from
the current study suggest that gender, sexual orientation, and
delinquent involvement are highly interconnected and attempts
to understand male–female differences in antisocial behavior
should also examine sexual orientation. Although it is unlikely
that the male–female gap in offending will be explained away
once sexual orientation is controlled, our results suggest that
including sexual orientation in a causal model might provide
new insight into how and why humans differ in their propensity
for offending.
Acknowledgments This research used data from Add Health, a program
project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mul-
lan Harris, and funded by a Grant P01-HD31921 from the Eunice Kennedy
Shriver National Institute of Child Health and Human Development, with
cooperative funding from 17 other agencies. Special acknowledgment is
due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the
original design. Persons interested in obtaining data files from Add Health
should contact Add Health, Carolina Population Center, 123 W. Franklin
Street, Chapel Hill, NC 27516-2524 (addhealth@unc.edu). No direct sup-
port was received from Grant P01-HD31921 for this analysis.
Appendix 1
See Table 6.
Arch Sex Behav (2016) 45:1759–1769 1767
123
Appendix 2
See Table 7.
References
Agnew, R. (1992). Foundation for a general strain theory of crime and
delinquency. Criminology, 30, 47–88.
Armour, S., & Haynie, D. L. (2007). Adolescent sexual debut and later
delinquency. Journal of Youth and Adolescence, 36, 141–152.
Barnes, J. C., Jorgensen, C., Beaver, K. M., Boutwell, B. B., & Wright, J.
P. (2015). Arrest prevalence in a national sample of adults: The role
of sex and race/ethnicity. American Journal of Criminal Justice,
40, 457–465.
Bell, N. J., O’Neal, K. K., Feng, D., & Schoenrock, C. J. (1999). Gender
and sexual risk. Sex Roles, 41, 313–332.
Bennett, S., Farrington, D. P., & Huesmann, L. R. (2005). Explaining
gender differences in crime and violence: The importance of social
cognitive skills. Aggression and Violent Behavior, 10, 263–288.
Biglan, A., Metzler, C. W., Wirt, R., Ary, D., Noell, J., Ochs, L., …Hood,
D. (1990). Social and behavioral factors associated with high-risk
sexual behavior among adolescents. Journal of Behavioral Med-
icine,13, 245–261.
Boutwell, B. B., Barnes, J. C., Deaton, R., & Beaver, K. M. (2013). On
the evolutionary origins of life-course persistent offending: A
theoretical scaffold for Moffitt’s developmental taxonomy. Jour-
nal of Theoretical Biology, 322, 72–80.
Cantor, R. J. (1982). Family correlates of male and female delinquency.
Criminology, 20, 149–167.
Ellis, L., Hoffman, H., & Burke, D. M. (1990). Sex, sexual orientation
and criminal and violent behavior. Personality and Individual
Differences, 11, 1207–1212.
Ellis, L.,& Walsh, A. (2000). Criminology: A global p erspective.Boston:
Allyn and Bacon.
Garnette, L., Irvine, A., Reyes, C., & Wilber, S. (2011). Lesbian, gay,
bisexual and transgender (LGBT) youth and the juvenile justice
system. In F. T. Sherman & F. H. Jacobs (Eds.), Juvenile justice:
Advancing research, policy, and practice (pp. 156–173). Hoboken,
NJ: Wiley.
Gottfredson, M., & Hirschi, T. (1990). A general theory of crime.
Stanford, CA: Stanford University Press.
Harden, K. P., Mendle, J., Hill, J. E., Turkheimer, E., & Emery, R. E.
(2007). Rethinking timing of first sex and delinquency. Journal of
Youth and Adolescence, 37, 373–385.
Harris, K. M. (2009). The National Longitudinal Study of Adolescent
Health (Add Health), Waves I & II, 1994–1996; Wave III, 2001–
2002; Wave IV, 2007–2009 [machine-readable data file and docu-
mentation].Chapel Hill, NC:Carolina Population Center,University
of North Carolina at Chapel Hill.
Harris, K. M., Florey, F., Tabor, J., Bearman, P. S., Jones, J., & Udry, R. J.
(2003). The National Longitudinal Study of Adolescent Health:
Research design [www document]. Retrieved from http://www.
cpc.unc.edu/projects/addhealth/design.
Haynie, D. L., & South, S. J. (2005). Residential mobility and adolescent
violence. Social Forces, 84, 361–374.
Koeppel, M. D. H. (2015). Assessing the association between self-
control and self-reported criminal behaviors between sexual orien-
tation groups. Criminal Justice Review, 40, 117–130.
Lippa, R. A. (2008). Sex differences and sexual orientation differences
in personality: Findings from the BBC internet survey. Archives of
Sexual Behavior, 37, 173–187.
Lussier, P., Proulx, J., & LeBlanc, M. (2005). Criminal propensity,
deviant sexual interests and criminal activity of sexual aggressors
against women: A comparison of explanatory models. Criminol-
ogy, 43, 249–282.
Mears, D. P., Cochran, J. C., & Beaver, K. M. (2013). Self-control theory
and nonlinear effects on offending. Journal of Quantitative Crim-
inology, 29, 447–476.
Metzler, C. W., Noell, J., Biglan, A., Ary, D., & Smolkowski, K. (1994).
The social context for risky sexual behavior among adolescents.
Journal of Behavioral Medicine, 17, 419–438.
Meyer, I. H. (1995). Minority stress and mental health in gay men.
Journal of Health and Social Behavior, 36, 38–56.
Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. A. (2001). Sex differences
in antisocial behavior: Conduct disorder, delinquency, and
violence in the Dunedin Longitudinal Study. Cambridge, UK:
Cambridge University Press.
Table 6 Post hoc group comparisons for males (Bonferroni method)
Wave 1 Wave 2 Wave 3 Wave 4
H vs. MH N N, T N, T
H vs. B
H vs. MG V
Hvs.G N,V,T V
MH vs. B
MH vs. MG T N, T
MH vs. G N, V, T N, V, T N, T N, T
B vs. MG T
Bvs.G N,V,T T
MG vs. G
Hheterosexual, MH mostly heterosexual, Bbisexual, MG mostly gay, G
gay, Nnonviolent delinquency, Vviolent delinquency, Ttotal composite
delinquency
Letters (N, V, or T) are included in cells where the two group means are
statistically different from each other at the p\.05 level (equal variances
not assumed) after Bonferroni correction
Table 7 Post hoc group comparisons for females (Bonferroni method)
Wave 1 Wave 2 Wave 3 Wave 4
H vs. MH N, V, T N, V, T N, T N, V, T
H vs. B N, V, T N, V, T N, V, T N, T
H vs. ML N, T N N
H vs. L V, T N, V, T V, T
MH vs. B V N
MH vs. ML
MH vs. L N, V, T
B vs. ML
B vs. L N, T V
ML vs. L
Hheterosexual, MH mostly heterosexual, Bbisexual, ML mostly lesbian,
Llesbian, Nnonviolent delinquency, Vviolent delinquency, Ttotal
composite delinquency
Letters (N, V, or T) are included in cells where the two group means are
statistically different from each other at the p\.05 level (equal variances
not assumed) after Bonferroni correction
1768 Arch Sex Behav (2016) 45:1759–1769
123
Nedelec, J. L., & Beaver, K. M. (2012). The association between sexual
behavior and antisocial behavior: Insights from an evolutionary
informed analysis. Journal of Contemporary Criminal Justice, 28,
329–345.
Pollock, W., Menard, S., Elliott, D. S., & Huizinga, D. H. (2015). It’s
official: Predictors of self-reported vs. officially recorded arrests.
Journal of Criminal Justice, 43, 69–79.
Ramrakha, S., Bell, M. L., Paul, C., Dickson, N., Moffitt, T. E., & Caspi,
A. (2007). Childhood behavior problems linked to sexual risk
taking in young adulthood: A birth cohort study. Journal of the
American Academy of Child and Adolescent Psychiatry, 46, 1272–
1279.
Rowe, D. C., Flannery, A. T., & Flannery, D. J. (1995). Sex differences in
crime: Do means and within-sex variation have similar causes?
Journal of Research in Crime and Delinquency, 32, 84–100.
Sandfort, T. G. (2005). Sexual orientation and gender: Stereotypes and
beyond. Archives of Sexual Behavior, 34, 595–611.
Savin-Williams, R. C., & Joyner, K. (2013). The dubious assessment of
gay, lesbian, and bisexual adolescents of Add Health. Archives of
Sexual Behavior, 43, 413–422.
Thornberry, T. P., & Krohn, M. D. (2000). The self-report method for
measuring delinquency and crime. Criminal Justice, 4, 33–83.
Udry, J. R. (2003). The National Longitudinal Study of Adolescent
Health (Add Health), Waves I and II, 1994–1996; Wave III, 2001–
2002 [machine readable data file and documentation]. Chapel Hill,
NC: Carolina Population Center, University of North Carolina at
Chapel Hill.
Widom, C. S. (1977). A methodology for studying non-institutionalized
psychopaths. Journal of Consulting and Clinical Psychology, 45,
674–683.
Wilson, J. W., & Herrnstein, R. J. (1985). Crime & human nature: The
definitive study of the causes of crime. New York: The Free Press.
Zuckerman, M., Bone, R. N., Neary, R., Mangelsdorff, D., & Brustman,
B. (1972). What is the sensation seeker? Personality trait and
experience correlates of the Sensation Seeking Scales. Journal of
Consulting and Clinical Psychology, 39, 308–321.
Zuckerman, M., Tushup, R., & Finner, S. (1976). Sexual attitude and
experience: Attitude and personality correlates and changes
produced by a course in sexuality. Journal of Consulting and
Clinical Psychology, 44, 7–19.
Arch Sex Behav (2016) 45:1759–1769 1769
123
A preview of this full-text is provided by Springer Nature.
Content available from Archives of Sexual Behavior
This content is subject to copyright. Terms and conditions apply.
