Reconceptualizing Adolescent Sexual Behavior

Text-only Preview

Reconceptualizing Adolescent Sexual Behavior:
Beyond Did They or Didn’t They?
By Daniel J. Whitaker, Kim S. Miller and Leslie F. Clark
yet had sex, and the number and types of
partners among adolescents who have.8
C o n t ex t : Adolescent sexual behavior is typically studied as a dichotomy: Adolescents have had
sex or they have not. Broadening this view would lead to a greater understanding of teenagers’

It classifies adolescents into five gro u p s ,
sexual behavior.
on the basis of their experiences or ex-
pectations with re g a rd to heterosexual ac-
M e t h o d s : I n t e rv i ew data from 907 high school students in Alabama, New Yo rk and Puerto Rico
tivity: those who have not had sex and
were used to examine the relationships between sexual ex p e rience and a va riety of social, psy-
have a low expectation that they will do
chological and behav i o ral va ri a bl e s. Four groups of teenagers are compared: those who did not
so in the next year (delayers), those who
anticipate initiating sex in the next year (delayers), those who anticipated initiating sex in the
have not had sex but have a high expec-
n ext year (anticipators), those who had had one sexual partner (singles) and those who had
tation that they will in the next year (an-
had two or more partners (multiples).
ticipators), those who have had sex one
R e s u l t s : Compared with delaye r s, anticipators reported more alcohol use and marijuana use;
time (one-timers), those who have had sex
poorer psychological health; riskier peer behaviors; and looser ties to fa m i l y, school and church.
m o re than once but with only one partner
S i m i l a rl y, multiples reported more alcohol and marijuana use, riskier peer behaviors and looser
(steadies) and those who have had sex
ties to family and school than singles. Risk behav i o r s, peer behav i o r s, family va ri a bl e s, and school
m o re than once and with two or more
and church invo l vement showed a linear trend across the four categories of sexual behav i o r.
partners (multiples).
Conclusions: The traditional sex–no sex dichotomy obscures differences among sexually in-
P revious re s e a rch indicates that com-
experienced teenagers and among adolescents who have had sex. Prevention efforts must be
p a red with delayers, anticipators engage
tailored to the specific needs of teenagers with differing sexual experiences and expectations,
in more precoital behaviors (kissing, touch-
and must address the social and psychological context in which sexual experiences occur.
i n g )9 and have less informational sup-
Family Planning Perspectives, 2000, 32(3):111–117
p o r t ;1 0 multiples begin sexual activity ear-
lier and use condoms less than one-timers
and steadies.11Other findings support the
conceptualized in a negative and pro b-
validity of the typology: Adolescents who
lematic context,” with the intent of pre-
The study of adolescent sexual be-
havior has been motivated larg e l y
anticipate having sex in the next six
by the health and social pro b l e m s
venting diseases and unplanned pre g-
months are more likely to do so than are
that may result when young people have
n a n c i e s .5 For these and other re a s o n s ,
those who do not expect to,12 and teenagers
u n p rotected sexual intercourse. Pre v e n-
adolescent sexuality is typically concep-
who have had multiple partners begin sex-
tion efforts aimed at meeting national
tualized and studied as a dichotomy: Ado-
ual activity earlier and use condoms less
health objectives have focused on delay-
lescents have had sex or they have not.
than those who have had only one.1 3
ing sexual onset among adolescents who
H o w e v e r, the dichotomous sex–no sex
For our study, we adapted Miller and
have not had sex and promoting condom
view does not take into account the psy-
colleagues’ typology by combining one-
use among adolescents who are sexually
chological and social context in which sex-
timers and steadies into one group of ado-
active. Although the proportions of ado-
ual behavior occurs—for example, such
lescents who had had one sex partner,
lescents who delay sexual onset and who
factors as whether an adolescent has had
whom we term singles. We focused on
use condoms have increased somewhat,1
one partner or many, how long the young
comparisons between groups of teenagers
a great deal of risky sexual behavior con-
people have known each other, whether
who had had sex (i.e., singles and multi-
tinues.2 As a result, teenagers experience
alcohol is used at the time of a sexual en-
ples) and between those who had not had
a large number of unplanned pre g n a n c i e s
counter and the age diff e rence between
sex (i.e., delayers and anticipators), be-
and sexually transmitted diseases, in-
p a r t n e r s .6 As a result, this view limits the
cause these comparisons are obscured by
cluding HIV.3
ability of programs, educators and others
the traditional sex–no sex dichotomy. In
Our understanding of adolescent sexu-
to prevent teenagers from engaging in
addition, we examined the linear tre n d
ality is limited, and improving that un-
risky behaviors.7
a c ross the four groups to better under-
derstanding promises to speed the pro g re s s
A broader conceptualization of adoles-
stand the association between the social,
t o w a rd meeting the nation’s public health
cent sexual experiences will improve the
psychological and behavioral variables
objectives. For theoretical, practical and po-
understanding of adolescent sexuality and
and teenagers’ level of sexual experience.
litical reasons, most re s e a rch has focused
aid in preventing risky sexual behaviors.
We analyzed dependent measures fro m
on examining the correlates of early sexu-
In this article, we examine social, psy-
the perspective that sexual risk behavior is
al initiation and condom use, rather than
chological and behavioral diff e re n c e s
on understanding adolescents’ sexual ex-
a c ross an expanded typology of adoles-
Daniel J. Whitaker is re s e a rch psychologist and Kim S.
periences. Among these reasons are the se-
cent sexual experience.
Miller is re s e a rch sociologist, both at the Centers for Dis-
ease Control and Prevention, Atlanta. Leslie F. Clark is
c recy surrounding sexual behavior, which
A typology advanced by Miller and col-
associate pro f e s s o r, Department of Health Behavior,
has hindered open communication about
leagues considers the readiness to engage
School of Public Health, University of Alabama at Bir-
s e x u a l i t y,4 and the fact that “sexuality is
in sex among adolescents who have not
Volume 32, Number 3, May/June 2000

determined by multiple factors at multiple
year (rated on a scale from one, indicating
ceived control was made up of five items
l e v e l s .1 4 For example, teenagers’ sexual ac-
that they were sure it would not happen,
( “I have little control over the things that
tivity or abstinence may be supported by
to five, indicating that they were sure it
happen to me”; “There is really no way I
various levels of factors—individual (e.g.,
would happen); and the number of part-
can solve some of the problems I have”;
intellect and drug use), peer (e.g., norms
ners sexually experienced teenagers had
“Sometimes I feel that I’m being pushed
and behavior), familial (e.g., parental mon-
had. Data for one or more items were miss-
around in life”; “There is little I can do to
itoring and socioeconomic status) and in-
ing for 13 participants, who thus could not
change many of the important things in
stitutional (e.g., school and churc h ) .1 5 Te e n-
be classified; our analyses are there f o re
my life”; and “I often feel helpless in deal-
agers who have had sex differ from those
based on data for 894 adolescents.
ing with the problems of life”). Positive fu-
who have not with respect to attitudes and
In all, 37% of the sample had never had
t u re outlook comprised four items (“What
b e l i e f s ,1 6 peer norms,1 7 alcohol and dru g
i n t e rcourse and rated their expectation for
happens to me in the future mostly de-
u s e ,1 8 p a rental factors,1 9 school involve-
having intercourse in the next year as less
pends on me”; “I can do just about any-
m e n t2 0 and church involvement.2 1 We ex-
than 50%; we categorized these adoles-
thing I really set my mind to do”; “My fu-
amined whether diff e rences for those vari-
cents as delayers. Another 22% had never
t u re is what I make of it”; and “I have gre a t
ables existed for typology groups within
had intercourse but rated their expecta-
faith in the future”). Hopelessness in-
the traditional sex–no sex dichotomy and
tions for doing so in the next year as 50%
cluded three items (“Sometimes I feel
linearly across the typology gro u p s .
or more; we considered this group antic-
there is nothing to look forward to in the
ipators. Some 13% of participants had had
f u t u re”; “I just live for today”; “It’s re a l l y
Methods and Variables
i n t e rcourse with only one partner (sin-
no use worrying about the future, because
Sample and Procedure
gles), while 27% had had sex with more
what will be will be”). Scales were formed
Participants were drawn from the Family
than one partner (multiples).
so that higher scores indicate higher self-
and Adolescent Risk Behavior and Com-
esteem, greater control, a more positive fu-
munication Study, a cross-sectional study
Individual-Level Factors
ture outlook and greater hopelessness.
of adolescent-mother pairs conducted in
Risk behaviors. The risk behaviors we ex-
Each of the final two psychological mea-
1993–1994 in Montgomery, Alabama; New
amined were similar to those discussed in
s u res—having a role model and being a
York City; and San Juan, Puerto Rico. Par-
earlier re s e a rc h .2 3 Adolescents re p o r t e d
role model—was measured with a single
ticipants were re c ruited through high
whether they had ever smoked cigare t t e s ,
yes-or-no question.
schools that had an overre p resentation of
whether they had ever used alcohol,
black or Puerto Rican adolescents. A more
whether they had had five or more drinks
Peer-Level Factors
complete description of the sample has
on a single occasion during the past 12
We assessed adolescents’ perceptions of
been published elsewhere .2 2
months (heavy alcohol use) and whether
peer group norms regarding having sex,
I n t e rested adolescents telephoned re-
they had ever used marijuana. They also
being pregnant or having gotten someone
s e a rchers and were screened for eligibil-
reported the number of physical fights
p regnant, using alcohol and having been
ity at that time. Eligible adolescents were
they had been in during the last 12
in jail. We computed the proportion of an
14–16 years old at enrollment, were in
months, how often they had carried a
adolescent’s close friends who had en-
grades 9–11 and had lived with their
weapon to school during the past 12
gaged in each behavior on the basis of the
mother and in the recruitment area for at
months (rated on a scale from one, indi-
participant’s reports of how many close
least the past 10 years. Eligible mothers
cating never, to five, indicating always)
friends he or she had and how many of
w e re the biological or adoptive mother or
and whether they had ever been held
those had engaged in each behavior.
stepmother of the adolescent. Of the 1,733
overnight in jail or a detention center.
students who provided screening infor-
Psychological factors. Six psychological
Family-Level Factors
mation, 1,124 were eligible; 982 (87%) of
factors were measured: self-esteem, per-
P a renting variables. Multiple aspects of
the eligible pairs were interviewed.
ceived control, future outlook, hopeless-
the mother’s parenting were examined
An interviewer matched by ethnicity
ness, whether the adolescent has a ro l e
f rom the reports of the adolescent and the
and gender to the participant conducted
model and whether he or she is a ro l e
mother: monitoring, closeness, commu-
separate interviews with the mother and
model. All of these measures were based
nication and parenting locus of contro l .
the adolescent. Mothers were interviewed
on the adolescent’s report; the first four
We focused on mothers rather than fathers
first whenever possible (91% of the pairs),
w e re developed from items taken fro m
because the adolescents in the study had
to ease the adolescents’ concerns that their
validated scales. The four scaled items
d i ffering amounts of contact with their fa-
responses would be discussed with their
were factor-analyzed.
thers (46% did not live with their father,
m o t h e r. Analyses of the interview data re-
First, we factor-analyzed 11 items fro m
whereas living with their mother was an
vealed that 907 of the 982 pairs met all el-
Coopersmith’s self-esteem scale2 4 that had
inclusion criterion for the study), and be-
igibility re q u i rements. We used both the
been presented to the adolescents as a sin-
cause the mothers’ responses were used
adolescents’ and the mothers’ re s p o n s e s
gle scale. Four of the items loaded onto the
in some of the measures.
for our analyses.
l a rgest factor, which accounted for 36% of
Monitoring re p resents the extent to
the variance. Those four items (“I wish I
which parents are aware of their child’s
w e re diff e rent”; “I often wish I were some-
b e h a v i o r. Four items, from the strict-
We used three items to classify adolescents
one else”; “I like the kind of person I am”;
ness/supervision scale,2 5 w e re used to as-
a c c o rding to our adapted typology: and “I am very happy the way I am”) were
sess the extent to which the mother knew
whether teenagers had ever had penile-
retained as a measure of self-esteem.
w h e re the adolescent went at night, what
vaginal intercourse; the extent to which
Next, we factor-analyzed 12 items that
the adolescent did with his or her fre e
teenagers who had not had penile-vaginal
had been presented to the adolescents as
time, where the adolescent went most af-
i n t e rcourse expected to do so in the next
a single scale; three factors emerged. Per-
ternoons after school and who the ado-
Family Planning Perspectives

Ta ble 1. Individual-level correlates of sexual activity, by teenage rs’ sexual ex p e r i e n c e, and p-values showing significance of various effe c t s
and comparisons, Family and Adolescent Risk Behavior and Communication Study, 1993–1994

Sexual experience
Sexual experience
Effect of
with gender
Ever smoked
Ever used alcohol
Used alcohol heavily
(≥5 drinks) in past year
Ever used marijuana
Ever held in jail
No. of times in fight
Carried a weapon to school
in past year
Has role model
Is role model
Perceived control
Positive future outlook
Notes: na=not applicable because comparison was not performed. Where the interaction between sexual experience and gender was significant, remaining comparisons were performed separately for fe-
males and males. Linear effect was tested using the Mantel-Haenszel chi-square for categorical data and a contrast with weights of –3, –1, +1 and +3 for the four sexual experience levels for the continu-
ous data. Scaled items are scored so that the higher the score, the greater the feelings of self-esteem, control, etc.
lescent’s friends were. Adolescents and
want to do”; “I feel in control when it
dicating not at all, to five, indicating very
mothers responded to these items, so we
comes to my son/daughter”; “I allow my
important); “How much do you like
computed two indices of monitoring
son/daughter to get away with things”).
school?” (choices ranged from one, signi-
(α=.68 for adolescents; α=.71 for mothers).
Negatively worded items were re v e r s e d ,
fying not at all, to four signifying a lot);
The index of mother-child closeness
and we summed the 10 items to form the
“How far would you like to go in school?”
was based on adolescents’ responses to
index of parental locus of control (α= .85).
(adolescents could choose from among
four items (“My mother and I are good
Family structure. We assessed four mea-
five responses, ranging from not caring if
friends”; “My mother and I are really close
s u res of the family’s stru c t u re. On the basis
they graduate from high school to want-
to one another”; “I trust my mother”; and
of the mother’s report, we assigned the
ing to graduate from high school, techni-
“My mother really loves me”). The items
family’s monthly income to one of seven
cal school, college, or graduate or pro f e s-
w e re summed to form the index of ma-
categories, ranging from less than $200 to
sional school); and “How far do you think
ternal closeness (α=.86).
$4,000 or more; we treated this variable as
you will actually go in school?” (with the
We used seven questions from Barnes
a continuous measure. Each parent’s ed-
same five possible answers). To assess
and Olson’s communication scale2 6 to con-
ucation, as reported by the mother, was
school performance, we asked the ado-
s t ruct mother-child communication indices.
classified as less than high school gradu-
lescents what their grade point average is,
Both adolescents and their mothers re-
ate, high school graduate or beyond a high
whether they had been suspended in the
sponded to the items (e.g., “My mother and
school degree. Finally, we assessed past year and whether they had ever been
I can talk about almost anything”; “When
whether the household was single- or
held back a grade.
I ask questions, I get honest answers fro m
dual-parent from the adolescent’s report
We asked two questions to assess re l i-
my mother”), so we summed each set of re-
of whether a biological father, adoptive fa-
giousness: how often the adolescents
sponses to form separate indices (α=.90 for
ther or stepfather was present.
attend religious services (four possible re-
adolescents; α=.85 for mothers).
sponses ranged from never to about once
P a rental locus of control is the degre e
Institutional-Level Factors
a week or more) and how important their
to which a parent feels in control of her
We also examined participants’ involve-
religious beliefs are to them (five choices
child’s behavior. We assessed 10 items,
ment with two extrafamilial institutions—
ranged from not at all to very). The ques-
which were adapted from an established
school and the church. For school in-
tions were conceptually similar, and
scale for parenting locus of contro l2 7 ( e . g . ,
volvement, we asked several questions:
although they were not highly corre l a t e d
“I find that sometimes my son/daughter
“How important is it to do well at school?”
(r=.34), we averaged them to form a
can get me to do things I really did not
(possible responses ranged from one, in-
single index.
Volume 32, Number 3, May/June 2000

Table 2. Percentage of respondents reporting various peer behaviors, by teenagers’ sexual experience, and p-values showing significance of
various effects and comparisons

Sexual experience
Sexual experience
Effect of
with gender
Ever had sex
Ever used alcohol
Ever pregnant/made
someone pregnant
Ever in jail
N o t e s : na=not applicable because comparison was not performed. Where the interaction between sexual experience and gender was significant, remaining comparisons were performed separately for
females and males. Linear effect was tested using the Mantel-Haenszel chi-square for categorical data and a contrast with weights of –3, –1, +1 and +3 for the four sexual experience levels for the contin-
uous data.
Analytic Plan
parison (cell weights, –3, –1, +1, +3) for con-
they were less likely to have a role model
We conducted the analyses in several
tinuous variables and the Mantel-Haenszel
(males only) and to be a role model for
steps. First, for each dependent measure ,
c h i - s q u a re for categorical variables.
someone else (Table 1). For peer norms,
we examined the overall effect of adoles-
students who expected to begin having
cents’ level of sexual experience and
sex soon were more likely than those who
whether that effect diff e red for female and
Sexual Experience and Gender
did not to say that their friends had en-
male teenagers. To determine gender dif-
Adolescents’ level of sexual experience
gaged in sex, used alcohol and been in jail
f e rences in the effects of sexual experience,
had a significant main effect on every risk
or a detention center (both genders).
we tested the interaction between sexual
behavior and every psychological variable
Anticipators also reported less pare n t a l
experience and gender by using analyses
except perceived control (for the sample
monitoring, less closeness with their par-
of variance for continuous dependent mea-
overall) and positive future outlook (Ta b l e
ents (both genders) and poorer commu-
s u res and log-linear analysis for categori-
1, page 113).* It also had a significant im-
nication with their parents than delayers.
cal dependent measures. If the interaction
pact on all peer behaviors (Table 2), all par-
F i n a l l y, they reported lower school en-
was significant, the remaining analyses for
enting variables (Table 3) and all school
joyment, lower expectations for school
that variable were conducted separately
and religion variables (Table 4, page 11 6 ) .
achievement, more suspensions fro m
for females and males; if not, gender was
In contrast, sexual experience had a sig-
school and less involvement in a re l i g i o u s
not considered further for that variable.
n i ficant effect on only one family stru c t u re
institution than their peers who did not
Next, for dependent variables that
variable: living in a single-parent house-
expect to initiate intercourse within the
showed a significant main effect for sexu-
hold (Table 3).
next year.
al experience or a significant interaction be-
The interaction between sexual experi-
In sum, compared with delayers, an-
tween sexual experience and gender, we
ence and gender was significant for seven
ticipators are clearly in a high-risk context
tested the hypotheses by comparing de-
variables: lifetime alcohol use, having a
that is consistent with their expectation of
layers and anticipators, comparing singles
role model and perceived control (Table 1);
pending sexual initiation.
and multiples, and testing the linear tre n d
having friends who had been involved in
a c ross the four groups. We used planned
a pregnancy and having friends who had
Singles vs. Multiples
comparisons (which use the error term fro m
ever in been in jail or a detention center
Depending on whether they had had one
the omnibus test) for continuous dependent
( Table 2); maternal closeness (Table 3); and
partner or more, sexually experienced ado-
variables and two-group chi-square tests
the importance of doing well in school
lescents diff e red on risk behaviors, peer
for categorical dependent variables. To test
( Table 4). There f o re, we analyzed these
norms, parenting variables and involve-
the linear trend, we used a planned com-
variables separately for females and males.
ment in school and church, but not on psy-
chological or family stru c t u re variables.
* T h e re were no diff e rences in age between delayers and
Delayers vs. Anticipators
C o m p a red with those who had had only
anticipators, or between one-timers, steadies and mul-
Students who had not had sex and did not
one partner (singles), adolescents who had
tiples (see re f e rence 8). However, when we combined the
expect to within the next year (delayers)
had at least two (multiples) reported more
one-timers and the steadies into one group who had had
one partner, that group (singles) was slightly younger
d i ff e red from those who expected to ini-
lifetime alcohol use (females only), heavy
than the multiples—15.4 vs. 15.6 years, on average (t
tiate sexual activity soon (anticipators) on
alcohol use, marijuana use, experience in
[360]=2.04, p=.04). Because there were age diff e re n c e s
some or all variables in every category ex-
jail or a detention center, and weapon car-
between singles and multiples, we included age as a co-
cept family stru c t u re. Compared with de-
rying. Multiples also were more likely than
variate in these omnibus analyses. Age had almost no
i n fluence on the sexual experience effects. The only vari-
layers, anticipators reported more ciga-
singles to have friends who had had sex,
able to be influenced was living in a single-parent house-
rette use, lifetime alcohol use (both
who had used alcohol and who had been
hold. When age was included, the sexual experience ef-
genders), heavy alcohol use, marijuana
in jail or a detention center (both genders);
fect changed from a significant level (p=.04) to a
use and weapon carrying. They also re-
young men who had had more than one
n o n s i g n i ficant level (p=.10). Because age had little eff e c t
on the results, none of the analyses reported include age
ported lower self-esteem, less control (fe-
partner were more likely to have a friend
as a covariate.
males only) and more hopelessness, and
who had gotten someone pregnant than
Family Planning Perspectives

w e re their peers who had had one partner. variable showed a significant linear tre n d .
to prevent risky behavior. The data show
Multiples reported less parental moni-
For the psychological factors (Table 1),
social, psychological and behavioral dif-
toring and closeness (females only) than
only perceived control and hopelessness
f e rences between groups of adolescents
singles, and their mothers reported less
showed significant linear trends. The
whom re s e a rchers and program planners
monitoring and a lower locus of contro l
t rend for perceived control diff e red by
typically group together: Sexually inex-
than did the mothers of singles. Finally,
gender: Females reported less contro l
perienced teenagers who do not expect to
students who had had multiple partners
a c ross higher levels of sexual experience,
have sex soon differ from those who do
rated school as less important (females
while the opposite was true for males.
with respect to risk behaviors, psycho-
only), liked school less, expected lower
Hopelessness increased with higher lev-
logical health, peer norms, parenting fac-
achievement, reported more suspensions
els of sexual experience.
tors and school or religious involvement;
and were more likely to have been held
F i n a l l y, one family stru c t u re variable—
except for psychological health, these fac-
back a grade than were those who had had
living in a single-parent household—
tors also distinguish young people who
one partner.
showed a significant linear trend. The pro-
have had one sexual partner from those
Thus, compared with singles, multiples
portion of participants living with only
who have had two or more.
reported higher risk not only with re g a rd
one parent rose significantly as the level
Our findings for risk behaviors, peer
to their sexual behavior but also with re-
of sexual experience increased.
norms, parenting factors, and involve-
g a rd to their peer groups, family and
It is notable that for several variables—
ment in school and religion are consistent
school involvement.
suspended in the past year, heavy alcohol
with the findings of re s e a rchers who have
use, marijuana use, being in jail, peer pre g-
used the dichotomous sex–no sex ap-
Linear Trends
nancies (males only) and peers in jail
proach and examined similar dependent
We expected to find higher scores across
(males only)—the linear effect does not
m e a s u res. However, our data expand
the levels of sexual experience for vari-
appear to fully explain the pattern of data.
these findings by revealing diff e re n c e s
ables that support greater sexual activity
S p e c i fic a l l y, the effect associated with hav-
that the sex–no sex dichotomy obscures.
(e.g., having peers who are sexually ac-
ing had multiple partners is greater than
Family stru c t u re was not related to sex-
tive), and lower scores for variables that
the linear trend would predict.
ual experience in our sample, although
support abstinence (e.g., parental moni-
other studies have shown a re l a t i o n s h i p .2 8
toring). The predicted trends were stro n g
No clear trend emerged for the psycho-
for risk behaviors (Table 1), peer behav-
Our results bear out our argument that a
logical variables. Delayers reported gre a t e r
iors (Table 2), parenting factors (Table 3)
b roader conceptualization of adolescent
psychological health than anticipators
and involvement with school or re l i g i o n
sexual experience is necessary to fully un-
(e.g., greater self-esteem, less hopeless-
( Table 4): Within these categories, every
derstand teenagers’ sexual behavior and
ness), but singles and multiples did not dif-
Table 3. Family-level correlates of sexual activity, by teenagers’ sexual experience, and p-values showing significance of various effects and

Sexual experience
Sexual experience
Effect of
with gender
Adolescent’s report
Mother’s report
(adolescent’s report)
Adolescent’s report
Mother’s report
Mother’s locus of control
Mother’s education
<high school
High school
>high school
Father’s education
<high school
High school
>high school
Single-parent household
Notes: na=not applicable because comparison was not performed. Where the interaction between sexual experience and gender was significant, remaining comparisons were performed separately for
females and males. Linear effect was tested using the Mantel-Haenszel chi-square for categorical data and a contrast with weights of –3, –1, +1 and +3 for the four sexual experience levels for the con-
tinuous data. Scaled items are scored so that the higher the score, the greater the level of monitoring, closeness, etc.
Volume 32, Number 3, May/June 2000

Table 4. Institutional-level correlates of sexual activity, by teenagers’ sexual experience, and p-values showing significance of various effects
and comparisons

Sexual experience
Sexual experience
Effect of
with gender
Suspended in past year
Ever held back a grade
Importance of doing well
Likes school
Desired achievement
Expected achievement
Grade point average
Religious involvement
Notes: na=not applicable because comparison was not performed. Where the interaction between sexual experience and gender was significant, remaining comparisons were performed separately for fe-
males and males. Linear effect was tested using the Mantel-Haenszel chi-square for categorical data and a contrast with weights of –3, –1, +1 and +3 for the four sexual experience levels for the continu-
ous data. Scaled items are scored so that the higher the score, the greater the desired achievement, religious involvement, etc. Grade point average was measured on a four-point scale.
f e r. Other work examining the re l a t i o n s h i p
m o re effective. This approach would be
connections to institutions such as school
between self-esteem and sexual behavior
similar to Prochaska’s stages-of-change
and religion that may motivate adoles-
among adolescents has had mixed re s u l t s :
m o d e l ,3 2 which has been used successful-
cents to delay sexual activity.
Some re s e a rchers have reported no re l a-
ly to modify behavior such as smoking,
The typical approach is to provide in-
t i o n s h i p ,2 9 and some have found diff e re n t
mammography screening and fat intake.3 3
formation and skills to the individual ado-
patterns for males and females.3 0
Abstinence-based messages about sex-
lescent. Although such information and
We found an intriguing interaction be-
ual behavior may be effective for adoles-
skills are necessary, they may not be suffi-
tween sexual experience and gender with
cents who do not see themselves as re a d y
cient for adolescents whose context sup-
re g a rd to psychological control: As sexu-
for sex (delayers), but not for adolescents
ports their having sex. Interventions are
al experience increased, males re p o r t e d
who do (anticipators). If anticipators ini-
needed that improve important pare n t i n g
g reater control, but females reported less.
tiate sex in a short time, as longitudinal
skills (e.g., monitoring and communica-
These relationships warrant further study. studies suggest they do,3 4 they may need
tion) or increase young people’s commit-
The data also reveal linear re l a t i o n s h i p s
messages that focus on the potential con-
ment to school. Such interventions should
between level of sexual experience and
sequences of having sex, peer pre s s u re
be carried out early—that is, with childre n
several types of variables: Greater sexual
and skills in safer-sex negotiation and con-
who have not reached adolescence. Re-
experience was associated with greater risk
dom use. Likewise, adolescents who are
ductions in sexual activity have been found
behaviors, riskier peer norms, poorer par-
sexually active may need tailored mes-
among youngsters who were involved in
enting and less involvement in school and
sages that address their sexual experi-
community service3 5 or who participated
religion. However, the relationship be-
ences. Adolescents who have sex in the
in a program to increase commitment to
tween certain factors and sexual experi-
context of a committed relationship may
school, if the program was implemented
ence does not appear to be strictly linear;
respond best to interventions that addre s s
b e f o re grades five and six.36 The idea of ad-
this finding also warrants further study.
relationship issues such as trust and com-
d ressing the social context as a way to pre-
mitment, or that involve their partner.
vent risk behavior is not new, but has not
Implications for Interventions
Teenagers who have multiple sex partners
been widely implemented.
The data have two primary implications
do so in an extremely high-risk context,
F rom a re s e a rch perspective, it will be
for re s e a rchers and health educators who
as evidenced by our data, and additional
important to examine adolescents’ chang-
either study adolescents or provide young
work is needed to determine what kinds
ing sexual experiences (e.g., how adoles-
people with services related to their sexu-
of interventions and messages will be
cents move from being delayers to being
al behavior: They should assess adolescent
most effective for them.
anticipators). Given the linear trends acro s s
sexual experience in greater detail, and
The second implication is that preven-
levels of sexual experience in our data, it
they should assess and address the social
tion programs and health educators
is tempting to conclude that the variables
and psychological context in which sexu-
should address the social and psycholog-
examined are causal. However, our data
al experiences occur.
ical context in which sexual behavior oc-
a re strictly a snapshot of the social, psy-
First, assessing adolescents’ sexual ex-
curs as part of their intervention or mes-
chological and behavioral context of ado-
periences in greater detail is necessary to tai-
sage. Our data clearly show that riskier
lescents with differing sexual experiences.
lor interventions and messages from health
sexual experiences occur in a unique so-
We do not know whether the dependent
educators. Prevention programs and mes-
cial and psychological context, which may
m e a s u res we analyzed are antecedents or
sages are often targeted to specific gro u p s
be evident even before teenagers begin to
consequences of particular sexual experi-
on the basis of gender, ethnicity or age.3 1
engage in sex, as illustrated by the com-
ences, nor do we know how those vari-
Ta rgeting messages to adolescents’ specif-
parisons between delayers and anticipa-
ables influence movement into new sexu-
ic sexual experiences should make those
tors. Interventions must better address fac-
al experiences. Longitudinal re s e a rch is
messages even more relevant and there f o re
tors such as peer norms, parenting and
needed to answer those questions.
Family Planning Perspectives

5 . D i M a u ro D, Sexuality Research in the United States: An
14 in a high-risk adolescent population, Family Planning
Assessment of the Social and Behavioral Sciences, New Yo r k :
Our study had several limitations. First, as
Perspectives, 1996, 28(1):13–18.
Social Science Research Council, 1995, p. 17.
noted, the data are cross-sectional; longi-
2 2 . Miller KS et al., Family communication about sex:
6 . Miller KS, Clark LF and Moore JS, Sexual initiation
tudinal data are needed to untangle cause
what are parents saying and are their adolescents lis-
with older male partners and subsequent HIV risk be-
and effect. A second limitation concerns the
tening? Family Planning Perspectives, 1998, 30(5):218–222
havior among female adolescents, Family Planning Per -
& 235.
sample. Although we re c ruited students at
spectives, 1997, 29(5):212–214.
t h ree sites, we did not randomly sample
23. Jessor R, Risk behavior in adolescence: a psychoso-
7. DiMauro D, 1995, op. cit. (see reference 5).
cial framework for understanding action, Journal of Ado -
adolescent-mother pairs. Participants vol-
lescent Health, 1991, 12(8):597–605.
8. Miller KS et al., Adolescent heterosexual experience:
u n t e e red for the study, which may have re-
a new typology, Journal of Adolescent Health, 1997, 20(3):
2 4 . Coopersmith S, The Antecedents of Self-Esteem, S a n
sulted in the inclusion of particularly mo-
Francisco: W. H. Freeman, 1967.
tivated, well-adjusted participants. Still, the
9. Ibid.
level of sexual behavior in the sample was
2 5 . S t e i n b e rg L et al., Impact of parenting practices on
adolescent achievement: authoritative parenting, school
1 0 . Miller KS, Whitaker DJ and Clark LF, Information-
considerable: More than one-quarter had
involvement, and encouragement to succeed, Child De -
al support diff e rences between delayers and anticipa-
had multiple sex partners.
velopment, 1992, 63(5):1266–1281.
tors, Atlanta: Centers for Disease Control and Pr e v e n-
A third limitation is that the data were
tion, 1999.
2 6 . Barnes HL and Olsen DH, Parent-adolescent com-
s e l f - reported, and thus contain all of the
munication and the circumplex model, Child Development,
11. Miller KS et al., 1997, op. cit. (see reference 8).
p roblems inherent to self-report measure s .
1985, 56(2):438–447.
1 2 . Stanton BF et al., Longitudinal stability and pre-
Given that we collected self-reports fro m
2 7 . Campis LK, Lyman RD and Prentice-Dunn S, The
dictability of sexual perceptions, intentions, and behav-
two sources—the adolescent and the
p a rental locus of control scale: development and vali-
iors among early adolescent African-Americans, J o u r n a l
mother—and that those sources led to
d a t i o n , Journal of Clinical Child Psychology, 1986, 15(3):
of Adolescent Health, 1996, 18(1):10–19.
similar conclusions, self-report pro b l e m s
1 3 . Santelli JS et al., Multiple sexual partners among U.S.
probably did not undermine the validity
2 8 . Hogan DP and Kitagawa EM, The impact of social
adolescents and young adults, Family Planning Perspec -
of our re s e a rch. Finally, the study was con-
status, family stru c t u re, and neighborhood on the fertil-
tives, 1998, 30(6):271–275.
ity of black adolescents, American Journal of Sociology, 1 9 8 5 ,
ducted only with youth who identified
1 4 . Jessor R and Jessor SL, P roblem Behavior and Psy -
90(4):825–855; and Flick LH, Paths to adolescent par-
themselves as heterosexual. We do not
chosocial Development: A Longitudinal Study of Youth, N e w
enthood: implications for prevention, Public Health Re -
know whether the findings would be the
York: Academic Press, 1977; and Perkins DF et al., An eco-
ports, 1986, 101(2):132–147.
logical risk-factor examination of adolescent sexual ac-
same for homosexual youth; in fact, we do
2 9 . Robinson RB and Frank DI, The relation between self-
tivity in three ethnic groups, Journal of Marriage and the
not know if and how the typology applies
esteem, sexual activity, and pre g n a n c y, A d o l e s c e n c e , 1 9 9 4 ,
Family, 1998, 60(3):660–673.
to homosexual youth at all.
1 5 . Small SA and Luster T, Adolescent sexual activity:
30. Kowaleski-Jones L and Mott FL, Sex, contraception
an ecological, risk-factor approach, Journal of Marriage
and childbearing among high-risk youth: do diff e re n t
and the Family, 1994, 56(1):181–192.
The typical dichotomizing of adolescents
factors influence males and females? Family Planning Per -
16. Basen-Engquist K and Parcel GS, Attitudes, norms,
spectives, 1998, 30(4):163–169.
into sexually experienced and not sexu-
and self-efficacy: a model of adolescents’ HIV- related sex-
ally experienced is limiting because it nar-
3 1 . Jemmott JB, Jemmott LS and Fong GT, Abstinence
ual risk behavior, Health Education Quarterly, 1992, 19(2):
and safer sex HIV risk reduction interventions for African
rows the range of sexual behavior. This,
American adolescents: a randomized controlled trial,
in turn, restricts our understanding of how
17. DiClemente RJ, Predictors of HIV-preventive sexu-
Journal of the American Medical Association, 1 9 9 8 ,
sexual behavior develops and our ability
al behavior in a high-risk adolescent population: the in-
279(19):1529–1536; and St. Lawrence JS et al., Cognitive-
fluence of perceived peer norms and sexual communi-
to prevent risky behavior. Focusing on
behavioral intervention to reduce African American ado-
cation on incarcerated adolescents’ consistent use of
lescents’ risk for HIV infection, Journal of Clinical and Con -
adolescents’ sexual experiences and the
condoms, Journal of Adolescent Health, 1991, 12(5):385–390;
sulting Psychology, 1995, 63(2):221–237.
social and psychological context in which
and Fisher JD, Misovich SJ and Fisher WD, Impact of per-
those experiences occur can improve ef-
3 2 . P rochaska JO, DiClemente CC and Norc ross JC, In
ceived social norms on adolescents’ AIDS-risk behavior
search of how people change: applications to addictive
forts to prevent risky sexual behaviors and
and prevention, in: DiClemente RJ, ed., Adolescents and
behaviors, American Psychologist, 1992, 47(9):1102–1114.
AIDS: A Generation in Jeopardy, Newbury Park, CA: Sage,
promote sexual health.
1992, pp. 117–136.
3 3 . Pallonen UE et al., A 2-year self-help smoking ces-
sation manual intervention among middle-aged Finnish
1 8 . Lowry R et al., Substance use and HIV- related sex-
men: an application of the transtheoretical model, P re -
ual behaviors among U.S. high school students: are they
1 . Trends in sexual risk behavior among high school stu-
ventive Medicine,1994, 23(4):507–514; Rakowski W et al.,
related? American Journal of Public Health, 1994, 84(7):
dents—United States, 1991–1997, Morbidity and Mortal -
Assessing elements of women’s decisions about mam-
ity Weekly Report, 1998, 47(36):749–752; and Ku L et al.,
m o g r a p h y, Health Psychology, 1992, 11 ( 2 ) : 111 – 118; and
Understanding changes in sexual activity among young
1 9 . Holtzman D and Rubinson R, Parent and peer com-
Greene GW et al., Stages of change for reducing dietary
m e t ropolitan men: 1979–1995, Family Planning Perspec -
munication effects on AIDS-related behavior among U.S.
fat to 30% of energy or less, Journal of the American Dietetic
tives, 1998, 30(6):256–262.
high school students, Family Planning Perspectives,1 9 9 5 ,
Association, 1994, 94(10):1105–1110.
2. The Alan Guttmacher Institute (AGI), Sex and Amer -
27(6):235–240 & 268; Metzler CW et al., The social con-
34. Miller KS et al., 1997, op. cit. (see reference 8); and
ica’s Teenagers, New York: AGI, 1994; and Trends in sex-
text for risky sexual behavior among adolescents, Jour -
Stanton BF et al., 1996, op. cit. (see reference 12).
ual risk behavior…, 1998, op. cit. (see reference 1).
nal of Behavioral Medicine, 1994, 17(4):419–438; Resnick
MD et al., Protecting adolescents from harm: findings
3 5 . O’Donnell L et al., The effectiveness of the Reach for
3. AGI, 1994, op. cit. (see reference 2); and Valleroy LA
f rom the National Longitudinal Study on Adolescent
Health Community Youth Service learning program in
et al., HIV infection in disadvantaged out-of-school
Health, Journal of the American Medical Association, 1997,
reducing early and unprotected sex among urban mid-
youth: prevalence for U.S. Job Corps entrants, 1990
dle school students, American Journal of Public Health, 1 9 9 8 ,
278(10):823–832; and Romer D et al., Social influences on
t h rough 1996, Journal of Acquired Immune Deficiency Syn -
the sexual behavior of youth at risk for HIV exposure ,
dromes and Human Retrovirology, 1998, 19(1):67–73.
American Journal of Public Health, 1994, 84(6):977–985.
3 6 . Hawkins JD et al., Preventing adolescent health-risk
4 . Institute of Medicine, The Hidden Epidemic: Confro n t i n g
behaviors by strengthening protection during childhood,
20. Resnick MD et al., 1997, op. cit. (see reference 19).
Sexually Transmitted Diseases,Washington, DC: Nation-
A rchives of Pediatric and Adolescent Medicine, 1 9 9 9 ,
al Academy Press, 1996.
2 1 . Mott FL et al., The determinants of first sex by age
Volume 32, Number 3, May/June 2000