Demographic, Biological, Psychological, and Social Dating Study

This study applied a biopsychosocial approach to examine the predictors of the initiation of sexual intercourse during adolescence. The sample included 157 boys and 268 girls, of whom 85.4% reported having had sexual intercourse by age 21 to 22. Participants were surveyed at 3 time points; once when they were 13 to 14 years old, again at ages 16 to 17 years, and again 6 years later. Proportional hazards regression was used to examine the relation of demographic, biological, psychological, and social predictors to the timing of first intercourse. In the biopsychosocial model, different factors significantly predicted timing of first intercourse for girls and boys. For girls, always-married parents and less dating alone predicted later timing of first intercourse. Earlier timing of first intercourse for boys was related to associating with peers with lower achievement orientation and greater importance of popularity.

Adolescence is a key time to study the development of sexuality and sexual behavior patterns because this is when most individuals first experience sexual intercourse. In 1992, for example, over 60% of never-married youth, ages 14 to 21, reported having had sexual intercourse (U.S. Department of Health and Human Services, 1995). However, despite being a normative event during adolescence, the exact timing of first intercourse raises a variety of issues for individuals. For example, girls who have first intercourse earlier than average (between the ages of 10 and 14) tend to have more sex partners, a higher likelihood of having sex with high-risk men (e.g., HIV positive), and a greater risk for sexually transmitted infections (STIs; Greenberg, Magder, & Aral, 1992). In contrast, girls who have first intercourse later than average tend to have lower self-esteem and poorer social relationships than their peers (Crockett, Bingham, Chopak, & Vicary, 1996). Thus, the timing of first sexual intercourse warrants particular attention when examining sexual development.

Sexual behavior, like any other behavior, is a result of multiple domains of influence, including biological, psychological, and social influences. The idea that behavior is a result of multiple domains of influence has been discussed by many scholars who examine how these domains relate to one another and to individual behaviors (Ford & Lerner, 1992; Magnusson, 1988; Petersen, 1987). Individuals exist in a social context with which they interact in a reciprocal process, both influencing and being influenced by social factors. Social processes often interact with biological and psychological factors to produce subsequent behavior of individuals (see review by Susman, 1997). Thus, incorporating biological, social, and psychological domains in a single model of development should provide a more comprehensive explanation of a particular behavior.

In her influential discussion of sexuality and adolescence, Chilman (1983) argued that the development of sexual behavior needs to be understood from a holistic perspective including physical, psychological, and social influences. Although many factors are being studied in current research on adolescent sexuality (e.g., demographic factors, Heaton & Jacobson, 1994; biological factors, Rowe, Rodgers, & Meseck-Bushey, 1989; and social factors, DiBlasio & Benda, 1992), few studies are comprehensive and include factors from each domain of influence. Researchers of sexual behavior have often focused on one or two domains of influence at one time (e.g., demographic factors, Miller & Bingham, 1989; or biological and social factors, Smith, 1989). The few studies using comprehensive, multidimensional models (Bingham, Miller, & Adams, 1990; Crockett et al., 1996) have demonstrated the importance of incorporating each domain of influence into models predicting sexual behavior. Thus, in this article we use a biopsychosocial perspective to examine the timing of first sexual intercourse during adolescence, in addition to biosocial and biopsychological models.

The models are tested using proportional hazards regression. This method addresses two limitations that exist for much of the body of research on timing of first intercourse. With only a few exceptions (Capaldi, Crosby, & Stoolmiller, 1996; Costa, Jessor, Donovan, & Fortenberry, 1995; Heaton & Jacobson, 1994; Meschke & Silbereisen, 1997), most researchers examining the timing of adolescents’ sexual behavior have used undifferentiated categories based on the relative timing of first intercourse as the dependent variable, that is, early versus late (e.g., Crockett et al., 1996; Tubman, Windle, & Windle, 1996). This approach has two major problems: (a) It artificially reduces the variance available to be explained, and (b) it makes it difficult to include those adolescents who are sexually inexperienced. Proportional hazards regression allows for both the use of the actual age of first intercourse as the dependent variable and the inclusion of persons who have not yet experienced intercourse (Allison, 1984). To allow readers to compare the two approaches, both are presented in this article; however, the primary focus is on the results of the proportional hazards regression. The following discussion highlights important predictors of sexual behavior from within the biological, psychological, and social domains.

PREDICTORS OF FIRST SEXUAL INTERCOURSE
Biological Influences

Biological maturation, or pubertal change, initiates the adolescent phase of life (Petersen, 1987). Pubertal development begins with the production of hormones that have a direct influence on sexual motivation by increasing libido (Smith, 1989; Udry & Billy, 1987) and also an indirect influence on sexual involvement by initiating change in physical appearance through the development of secondary sex characteristics (Brooks-Gunn & Graber, 1994; Smith 1989). Given both of these types of influence, one would expect individuals who reach puberty at younger ages to also participate in sexual behavior (including intercourse) at younger ages.

The influence of pubertal changes on sexual behavior have been found to vary by gender, with pubertal development emerging as a better predictor of sexual behavior for boys than girls (Crockett et al., 1996; Udry & Billy, 1987). Social factors, such as parental control and gender role expectations, could explain these differences. For example, although adult physical appearance, such as breast or beard growth, may indicate to potential partners that an individual is sexually mature despite his or her chronological age, greater social constraints on female sexual behavior may decrease the predictive power of such biological development for girls (Petersen, 1987). Social mores regarding the appropriate age to participate in sexual activity and an individual’s value system may limit sexual behavior regardless of the person’s physical maturation. Thus, pubertal timing precedes psychological and social factors in the models. When the social measures are significant predictors of first intercourse, the moderation of the relation between pubertal timing and first intercourse by social factors is also considered.

Psychological Influences

Psychological factors have been included in models predicting the timing of first intercourse via a variety of ways, including attitudes, values, and emotional adjustment (Miller, Norton, Fan, & Christopherson, 1998; Tubman et al., 1996). Stack (1994) implied that premarital sexual behavior may be used as an aid when developing friendships, particularly when adolescents move repeatedly, as sex may serve as a way to combat loneliness. Evidence also suggests that both girls and boys have sex even when they do not want to in response to peer pressure or desires to be popular (Erickson & Rapkin, 1991; Muehlenhard & Cook, 1988). Thus, values placed on popularity appear to encourage decisions to participate in sexual behavior. If adolescents hold these values and perceive sexual behavior as an important way to become popular, they may participate in sexual activity earlier than adolescents who hold alternative values.

Emotional adjustment may also influence whether or not an individual participates in sexual activity. For example, higher levels of depression have been found to predict earlier initiation of sexual intercourse (Tubman et al., 1996). Sexual activity may serve as a coping mechanism, or a way for adolescents to temporarily “feel better” about themselves or situations. Thus, the importance of popularity to the adolescent and depressed mood are included as psychological factors in the models.

Social Influences

Social factors have also been emphasized in studies of the timing of first intercourse. Several researchers (Petersen, 1987; Smith, 1989; Udry, 1990) have suggested that sexual behavior may be influenced more by social constraints than biological influences. A person may experience an increase in libido, but not act on it due to established norms of sexual behavior. Of particular interest are the social realms of family and peers.

The nature of family processes during adolescence may increase or decrease an adolescent’s likelihood of participating in sexual activity. Parental control and regulation has been quite consistently related to adolescent sexual behavior (see review by Miller, 1998). Higher levels of parental supervision and control have been associated with a later age of first intercourse for both boys and girls (Hogan & Kitagawa, 1985; Ku, Sonenstein, & Pleck, 1993). In addition to adolescents’ perceptions of overall parental control in their lives, curfews provide a more specific avenue by which parents may attempt to better supervise their adolescent’s dating behavior. Although having a curfew in and of itself will not prevent sexual activity, it may limit the opportunity adolescents have for sexual activity.

Peers are another source of information and influence particularly regarding sexual behavior. Zelnik (1983) found that most 15- to 17-year-old sexually experienced girls (68.7%) had their first intercourse in an engaged or steady relationship. As expected, having a boyfriend or girlfriend is a significant predictor of adolescent sexual activity (Scott-Jones & White, 1990). Indeed, persons who date at an earlier age also have steady relationships at an earlier age and report being more sexually active during their late teen years (Thornton, 1990). With these findings in mind, the frequency with which an adolescent is dating alone in 7th grade (unchaperoned and without other peers) may also be related to an earlier age of first intercourse.

Peer groups also furnish information about sexual attitudes, act as role models of sexual behavior, and provide opportunities for adolescents to participate in sexual activity (Smith, 1989). Peers have been shown to be more influential than parents in predicting earlier initiation of sexual intercourse (Costa et al., 1995; Wyatt, 1989). School is a context in which adolescents develop and maintain friendships and peer groups. When students are tracked at a certain academic level, they spend a great deal of time in classes with the same students during a particular school day. Peer group associations may develop during classes as a result of this.

Fuligni, Eccles, and Barber (1995), examining the data used in this article, reported that ability groupings were significantly related to an individual’s association with achievement-oriented peers. Persons in higher ability tracks reported having higher achievement-oriented peers. This was true despite the fact that neither the individual’s ability level alone nor the interaction between ability and ability tracks had a significant effect on peer achievement orientation. In other words, school track, independent of one’s ability, was strongly associated with the characteristics of one’s peers, particularly achievement orientation, which, in turn, has been shown to influence adolescent sexual behavior (Grunbaum & Basen-Engquist, 1993; Scott-Jones & White, 1990). Therefore, school educational track reflects the extent to which one affiliates with achievement-oriented peers. The social measures in the models include parental control, having a weekend curfew, dating alone (vs. group dating), and peer achievement.

Demographic Variables

Additionally, we are interested in the predictive power of a biopsychosocial model above and beyond the contribution of those demographic factors shown to predict the onset of adolescent sexual behavior (DiBlasio & Benda, 1992). Therefore, demographic factors (married parents and maternal education) were included in all models. Adolescents growing up in intact families (Newcomer & Udry, 1987) and having highly educated mothers (Heaton & Jacobson, 1994) have reported later timing of first sexual activity. Finally, in recognition of gender differences and influences on behavior, such as the association between pubertal development and timing of first intercourse, each model is examined separately by gender.

Goals of this study. This study has two goals. First, differences in biological, psychological, and social factors are examined by groups based on the timing of first intercourse. Second, the timing of first sexual intercourse is examined by applying a variety of models. Specifically, four different models were tested: biological, biopsychological, biosocial, and biopsychosocial. It was expected that the full biopsychosocial model would significantly predict the timing of first intercourse above and beyond the prediction of the other three models.

METHOD
Sample

The data used in this study come from Waves 4, 5, and 7 of the Michigan Study of Adolescent Life Transitions (MSALT; for full sample details, see Eccles et al., 1989). MSALT began in 1983 when the respondents were in 6th grade in 10 school districts in southeastern Michigan. Wave 4 data were collected in 1985 when the respondents were in Grade 7, ages 13 to 14. Wave 5 data were collected in 1988 when the respondents were approximately 16 and 17 years old. Wave 7 data were collected in 1992 and 1993 when the respondents were about 21 and 22 years old.

Data from Waves 4 and 5 were used to examine the proximal correlates of first sexual intercourse. In this sample, one person (less than 1% of the sample) reported having sexual intercourse prior to Wave 4 and approximately 11.5% of the respondents reported having first intercourse prior to Wave 5. Retrospective reports from Wave 7 data were used to estimate the age at which first sexual intercourse occurred. Requiring that respondents participated in Waves 4, 5, and 7 yielded a subsample of 170 boys and 299 girls. The sample is primarily middle or working class, from small urban or suburban communities, and White (over 90%). Race has been found to be a significant correlate of the timing of first intercourse (The Alan Guttmacher Institute [AGI], 1994); however, the unequal distribution of the respondents by race does not allow comparisons by race to be made.[1] Thus, only White respondents were included in the analyses. This restriction limited the sample to 157 boys and 268 girls. Of this subsample, 85.4% reported having had voluntary sexual intercourse at Wave 7.

Analyses of Sample Bias

Two issues are likely to have contributed to a biased study sample: (a) Respondents were required to have data for each of the measures for three waves of data collection, and (b) sample bias often occurs in longitudinal samples. Specifically, there was a concern that those persons included in this study may have differed from the excluded (nonstudy) respondents on factors associated with this study. A base sample was created to determine whether there was a difference between the participants who had responded to the surveys at Waves 4, 5, and 7 (study sample) and those persons who reported about pubertal timing at Wave 5 but were missing data from Waves 4, 7, or both. Wave 5 was chosen as the base sample because the funding for MSALT was restricted during this wave of data collection, making it impossible to collect data for all original school districts. Thus, some earlier respondents were excluded without any choice on their part. The Wave 5 base sample included 1,124 persons (628 girls and 496 boys), of which 37.8% (425) were in the study sample. From the nonstudy portion of the base sample, 178 persons (92 girls and 86 boys) were missing at Wave 4. An additional 467 (230 girls and 237 boys) were missing data at Wave 7. Ten of the respondents were excluded from the study sample, as they did not report the educational status of their mother and 44 were excluded due to the race variable. Specifically, 17 girls and 7 boys were not White and 20 respondents did not report their race.

The analyses were divided into two parts, depending on whether the variables examined were categorical or continuous. The measures included in the analyses reflect those used in the study. These measures are described in the “Measures” section. The categorical variables included gender, maternal education, and married parents. The continuous variables included pubertal timing (grade), importance of popularity, depressed mood, alone on a date, curfew on weekend, parental control, peer achievement, and age of first intercourse (if experienced).

In examining the categorical set of variables, the study sample was comprised of a higher percentage of girls, 63.1%, X2(1,N = 1, 124) = 14.32, p < .001 than the attrition sample (51.5% female). No significant differences were found in the distribution of college-educated mothers or married parents by sample membership. T tests were used to examine potential differences in the mean levels of continuous measures by sample. The study sample had significantly higher levels of depressed mood (M = 2.80), t = -2.46, p < .05, and peer achievement (M = 2.75), t = -5.96, p < .001, than the nonstudy sample (M = 2.66 and M = 2.37, respectively). The study sample also reported less dating alone (M = 1.66), t = 4.36, p < .001, than the nonstudy sample (M = 1.90). No differences between the groups were found for pubertal timing, parental control, importance of popularity, weekend curfew, or age of first intercourse.

In sum, 4 of the 10 variables of interest differed significantly between the attrition and the study samples. It appears that the study sample may have had advantages over the attrition sample that would delay their timing of intercourse: higher percentage of girls, more association with high-achieving peers, and less dating alone. Yet the study sample also reported higher levels of depressed mood, which has been associated with earlier initiation. However, despite the differences for these predictors, it is important to note that no mean differences were found for the critical variable of interest: the age of first intercourse.

Measures

The dependent variable was age of first voluntary sexual intercourse. This construct was measured by the question, "Have you ever voluntarily had sexual intercourse? If yes, how old were you the first time (in years)?" Mixed findings exist about using retrospective reports of the timing of first intercourse. Newcomer and Udry (1988) found that only 7% gave a false report of first intercourse when they were asked initially. Alexander, Somerfield, Ensminger, Johnson, and Kim (1993) found that 67% of their respondents were inconsistent in reports of first sexual intercourse. Both studies focused on junior high and early high school reports. Further MSALT data are not available to test the reliability of the dependent measure.

Two versions of the timing of first intercourse were used in the analyses. The first, Sex Group (a three-level categorical variable) was used for descriptive analyses: Earlier (intercourse before age 16), Later (intercourse at or after age 16), and Inexperienced. The second was designed for proportional hazards regression: If the person had had voluntary intercourse, then the variable value was the age at which first intercourse occurred; if the person was sexually inexperienced, then the variable was the respondent's age at the time he or she filled out the Wave 7 questionnaire. Of the persons who reported being sexually experienced at Wave 7 (n = 371), no difference was found in the age of first voluntary intercourse between boys (M = 16.27, SD = 1.96) and girls (M = 16.24, SD = 1.75), t(363) = -.15, p > .10.

The predictors, taken primarily from the Wave 4 data, represent four categories: demographic, biological, psychological, and social predictors. The two demographic predictors were parental marital status and maternal education. Parental marital status from Wave 7 was based on the question, “Are your parents currently married and living together?” (1 = yes, 0 = no). Of the 285 respondents stating “yes,” over 95% reported that this status was true for 15 years or longer. Thus, a value of 1 refers to currently married parents and 0 is other. Maternal education (Wave 5) was the educational level reported by the adolescent for his or her mother. The education measure was restructured into three levels: 1 (less than high school), 2 (high school graduate), and 3 (some college). Over 55% of the sample reported that their mothers had some college education at Wave 5. No difference emerged by gender: girls (M = 2.38, SD = .72); boys (M = 2.51, SD = .72), t(425) = -1.73, p < .10.

The biological predictor, measured at Wave 5, was a retrospective report of pubertal status. Boys responded to the question, “Kids your age grow at different rates, but usually everyone has a time when they grew faster than at other ages. Has this happened to you yet? If yes, what grade were you in when this happened?” and girls answered, “Have you had your first period yet? If yes, when did you get it?” The answers ranged from Grades 5 to 10, with a mean response value of 6.82 (n = 268) for girls and 7.64 (n = 157) for boys. Two of the girls had not experienced menarche by Wave 5 and were dropped from the analyses.

Given the retrospective nature of this measure, steps were taken to support its reliability. Respondents had reported on identical measures of pubertal timing in both Waves 5 (10th grade) and 6 (12th grade). Correlation analyses confirmed that a significant correlation between the two waves of data occurred for both boys (alpha =.72; r = .56; p < .001; n = 62) and girls (alpha = .57; r = .42; p < .001; n = 119). Furthermore, almost half of the respondents (48.6%) reported the same grade of pubertal development for both years, and an additional 32.0% were within 1 year when comparing the measures from Waves 5 and 6.

Two predictors were included as psychological measures. Alpha levels reported refer to the subsample of MSALT participants used in this study. The scale score for each respondent was the unit-weight mean value of all the items included in the scale. Each item was commonly answered on a 7-point Likert-type scale.

Importance of Popularity, Wave 4 (alpha =.72), included two items: “For me, being popular is …” and “How important is it to you to be popular with girls (boys)?”; 1 (not at all important) to 7 (very important).

Depressed Mood, Wave 4 (alpha = .75), was measured with three items: “How often do you lose your appetite or eat a lot when upset?” “How often do you feel unhappy, sad, or depressed?” and “How often do you feel lonely?”; 1 (never) to 5 (very often).

Four social context predictors were examined. Again, the respondent’s scale score was the mean value for the variables comprising each scale and each item was answered on a 4- or 7-point Likert-type scale.

Alone on a Date, Wave 4, was based on one question: “How often do you go out with a boy (girl) alone?”; 1 (never) to 4 (very often).

Curfew on Weekend, Wave 4, was also based on one question: “When you’re out with your friends on a weekend night do your parents set a time you have to be home?”; 1 (never) to 7 (always).

Parental Control, Wave 4 (alpha = .73), included seven items: “My parents want me to follow their directions even if I disagree with their reasons,” “My parents worry that I am up to something they won’t like,” “I have to ask my parents for permission to do most things,” “My parents get upset if I disagree with them when their friends are around,” “I do not know why I am supposed to do what my parents tell me to do,” “I have a lot of fights with my parents about their rules and decisions for me,” and “My parents treat me more like a little kid than like an adult”; 1 (never true) to 4 (always true).

Peer Achievement, Wave 5, was based on ninth-grade school records of enrollment in 4 levels of mathematics courses: 1 (low), 2 (regular), 3 (college prep), and 4 (honors track). (For further explanation of this measure, see Updegraff, Eccles, Barber, & O’Brien, 1996.)

RESULTS
Descriptive Analyses

All analyses were conducted using the SPSS statistical package. Prior to conducting the primary analyses, correlation matrices of the predictors and dependent variable (whether or not the person had had sex) by gender were analyzed. In examining the correlation matrix, particular attention was given to the correlation values between the various predictors. A high correlation value between the predictors increases the likelihood of multicollinearity. The correlation coefficients ranged from -.33 to .37, thus issues of multicollinearity were dismissed (Affifi & Clark, 1990).

An analysis of variance (ANOVA) was used to determine whether the mean level of the demographic, biological, psychological, and social predictors differed depending on if and when the persons reported first intercourse. The three-level variable, Sex Group (Inexperienced; Later, 16 or older; and Earlier, before age 16) was used in these analyses. If significant, follow-up Tukey tests were used to determine which of the Sex Groups differed on the various characteristics. The ANOVAs were analyzed separately by gender.

Table 1 summarizes the differences between the female sex groups in mean levels of the demographic, biological, psychological, and sociological predictors: multivariate analysis of variance (MANOVA), A = .75; F = 4.49, p < .001. Regarding the demographic variables, the inexperienced group reported a higher mean level of living with married parents than the earlier group. In other words, a greater proportion of the inexperienced group had married parents compared to their earlier timing peers. The later initiating and inexperienced groups also reported higher levels of maternal education than the earlier timing group. Female early initiators reported earlier menarche than the later initiating and inexperienced groups.

The earlier experienced group placed greater value on popularity than the inexperienced group. The level of depressed mood was found to be significantly higher for the early initiators than the later initiators. In reference to the social context variables, the early initiators reported higher frequencies of being alone on a date than the later initiators, who in turn reported more frequent dating alone than the sexually inexperienced group. The sexually inexperienced group reported more association with high-achieving peers than the early and late initiators. No significant contrasts for the girls were found for curfew or parental control.

The overall model was significant for boys: MANOVA, A =.76, F = 2.41, p < .01. Compared to the girls, fewer significant contrasts emerged for the boys (see Table 2). The model for growth spurt was significant, but the contrast between the early initiators and the inexperienced groups was only approaching significance, p <10. Regarding the importance of popularity, both sexually experienced groups reported greater importance than the sexually inexperienced group. The early initiators reported dating alone significantly more than the later initiators and the inexperienced group. Finally, the inexperienced group reported greater association with high-achieving peers than did the early initiators. No significant contrast emerged for parental marital status, maternal education, depressed mood, weekend curfew, or parental control for boys.

Analysis of Age at First Voluntary Intercourse

Event history analysis was utilized to examine the timing of intercourse and its association with the demographic, biological, psychological, and social predictors. Specifically, Cox’s (1972) proportional hazards regression, which allows the inclusion of categorical and continuous predictors, was used to address this issue.

A hazard rate, or the instantaneous risk that the event (first intercourse) will occur at a given moment if the event has not occurred before this time (Yamaguchi, 1991), is calculated in event history analysis. The results of proportional hazards regression are interpreted using risk ratios based on the parameter estimates. A risk ratio exceeding 1.0 means that as the unit value of the predictor increases, the hazard rate also increases, that is, earlier timing of the event. If the risk ratio is less than 1.0, then an increase in the predictor value would be related to later timing of the event, or a decrease in the hazard rate (SAS Institute, 1990). For example, if the importance of friends has a risk ratio of 1.75 regarding the timing of intercourse, for each unit of increase in the importance of friends, persons who have not yet experienced sexual intercourse would increase their hazard rate, or likelihood of having sex, by 75%.

Using proportional hazards regression, four models predicting the timing of first sexual intercourse were analyzed. All four models were examined separately for boys and girls. The first model included two demographic predictors (parental marital status and maternal education) and the biological predictor (pubertal timing). In comparing this model to the subsequent models, we determined if the inclusion of the biological, psychological, and social predictors resulted in a decrease in the effect of the demographic predictors on the timing of first intercourse and whether an increase in the prediction power of the model occurred. The latter was detected by calculating whether there was a significant increase in the chi-squared value of the expanded model.

The second model included psychological predictors (popularity and depressed mood), in addition to the demographic and biological predictors. The third model considered the effects of demographic, biological, and social predictors (e.g., curfew and dating) on timing of first intercourse. The final model included demographic, biological, psychological, and social predictors of first intercourse. All of these models are summarized in Table 3.

The first regression model included demographic and biological predictors and was significant only for girls. Having married parents, a more educated mother, and later pubertal timing delayed the girls’ first intercourse.

The second model included demographic, biological, and psychological predictors. The overall model was significant for boys and girls. Compared to the first model, a significant increase in the chi-squared value for this model emerged for the girls, x2(2,N = 268) = 6.5, p < .05, and the boys, X2(2,N = 157) = 8.49, p < .01. Married parents and pubertal timing remained significant predictors for the girls. More importance placed on popularity predicted earlier intercourse for both girls and boys. Later pubertal timing also predicted a delay in first intercourse for the boys.

The third model (biosocial) was significant for boys and girls. Compared to the first model, a significant increase in this model’s chi-squared value was found for girls X2(4,N = 268) = 25.50, p< .001, and was approaching significance for the boys, X2(4,N = 157) = 9.36, p < .10. In this model, having married parents remained a significant demographic predictor for the girls. Pubertal timing approached significance for the boys. Higher frequency of dating alone predicted earlier timing of intercourse for girls. For the boys, associating with higher achieving peers predicted later timing.

Social factors included in the model may be working in conjunction with biological factors to produce sexual behavior (Katchadourian, 1990). To test this idea, the moderating effects of social characteristics on the relation between pubertal development and the timing of first intercourse were tested using interaction terms in the regression. None of the social predictors were found to moderate the relation between pubertal timing and first intercourse.

The final model included demographic, biological, psychological, and social contextual factors. Compared to the biopsychological model, the biopsychosocial model resulted in a significant increase in the model’s chi-squared value for girls, X2(4,N = 268) = 23.54, p < .001, and approached significance for the boys, X2(4,N = 157) = 8.46, p < .10. For boys, a significant increase in the chi-squared value was found in comparing the biosocial and the biopsychosocial models, X2(2,N = 157) = 7.59, p < .05. No significant differences between the chi-squared values of the biosocial and biopsychosocial models emerged for girls. A comparison was also made between the biopsychosocial model and the initial biological model. Here, the model chi-squared value increased significantly for both boys, X2(7,N = 157) = 16.95, p < .01, and girls, X2(7,N = 268) = 30.04, p < .001.

In the biopsychosocial model, girls with always-married parents reported later timing of intercourse. Dating alone remained a significant predictor for the girls. For the boys, less importance placed on popularity with peers and having higher achieving peers continued to be significantly related to a delay of first intercourse.

DISCUSSION
This study has examined the association of demographic, biological, psychological, and social predictors with the timing of initiation of sexual intercourse. Varying patterns of predictors emerged for boys and girls across the different models. For example, although the biopsychosocial model of first intercourse was significant for both boys and girls, pubertal timing and dating alone were significant predictors for the girls, whereas for the boys, the importance of popularity and achievement level of peers were significant. This study also examined biological, biopsychological, and biosocial models. This discussion focuses on the varying impact of demographic, biological, psychological, and social predictors across models and offers possible explanations for these contrasts.

The demographic measures emerged as an important set of predictors for the girls. In the biological model, having married parents and more highly educated mothers were related to later first intercourse for girls. The effect of having married parents on adolescent female sexual behavior is similar to that found by Newcomer and Udry (1987), who also reported this effect only for girls. Having fewer than two parents has been associated with different levels of various family processes (Barber & Eccles, 1992). However, despite the inclusion of parental control in the later models, having married parents remained a significant predictor for the girls. It would appear that the relevant family processes (like communication or monitoring) that might moderate or mediate the relation between parental marital status and first intercourse are not represented in the selected predictors.

Regarding the other demographic predictor, having a highly educated mother also has been associated with a delay in adolescent sexual behavior (Heaton & Jacobson, 1994; Leigh Weddle, & Loewen, 1988). Mothers who have prioritized education for themselves could be particularly salient role models for daughters. This form of support could lead these daughters to prioritize education and delay intercourse. However, this predictor was no longer significant with the addition of the psychological and social factors.

Similar to previous studies (Flannery, Rowe, & Gully, 1993; Udry, 1979; Westney, Jenkins, & Benjamin, 1983), the biological predictor, pubertal timing, was influential for both boys and girls. Earlier research (Udry & Billy, 1987) has concluded that for girls but not boys, social controls play a more important role than biological factors in the initiation of sexual activity. In this study, however, the addition of social predictors (biosocial and biopsychosocial models) decreased the contribution of pubertal timing in predicting first coitus for both boys and girls. Analyses were conducted to examine the possible moderating effects of the social factors on the relation between pubertal timing and first intercourse. No significant moderating effects were found. However, this does not preclude the possibility that together the social predictors might buffer the effect of pubertal timing on first intercourse. Alternately, it is also possible that this study is only partially addressing influential social factors. Future studies could examine social predictors other than those included in this article. The fact that pubertal timing was measured differently by gender might also be affecting the results. Given the data limitations, potential differences could not be tested.

The psychological predictors emerged as a strong set of predictors. Compared to the biological model, the biopsychological model produced a significant increase in the model’s chi-square value for boys and girls. As predicted, placing greater importance on popularity was related to earlier timing. Sexual behavior was hypothesized to serve as a potential tool to increase popularity. However, in the biopsychosocial model, although the importance of popularity was approaching significance as a predictor of earlier intercourse for the boys, it was no longer significant for the girls. This difference in the significance of popularity as a predictor by gender could be due to the social perceptions of sex that vary by gender. One female adolescent in Orenstein’s book, School Girls (1994), stated, “If a girl does it with a guy just because she thinks he’s cute, she’s a ho’, but if men do it with young women ’cause they think the girl is cute, then they’re cool” (p. 220). In other words, compared to girls, social stigmas allow sex to serve boys as a more viable and effective method by which to gain social status.

Social predictors appeared to be especially relevant for girls, whose models resulted in a significant increase in the chi-squared value between the biological and the biosocial model. Dating alone predicted earlier first intercourse for girls; however, this measure only approached significance for the boys. This lack of a significant effect for boys was possibly due to the smaller sample size for the boys, resulting in a model with less power.

With potentially lower statistical power, it is quite striking that only boys associating with high achieving peers reported a delay in intercourse and that this predictor remained significant even after the psychological predictors were added to the biosocial model to form the biopsychosocial model. Three post hoc explanations may aid in clarifying this finding. First, the respondents’ associations with high-achieving peers are likely to reflect their values of achievement. Placing a higher value on education has been shown to deter timing of first intercourse (Miller & Sneesby, 1988; Moore, Peterson, & Furstenburg, 1985; Scott-Jones & White, 1990). Second, being in a group of higher achieving peers may buffer one from the problems of declining self-esteem associated with school transition (Lord, Eccles, & McCarthy, 1994). Low levels of self-esteem, in turn, have been associated with earlier timing of first intercourse (Miller & Fox, 1987). Focusing on academic achievement may serve as a distraction from the many changes associated with a school transition, thus decreasing the risk of a decline in self-esteem and delaying first intercourse. Third, association with achievement-oriented peers could reinforce a focus on educational goals rather than hetero-social goals. This could lead students with higher achieving peers to initiate intercourse later than their peers with less achievement-oriented classmates. All three are likely to occur and may actually serve to reinforce one another.

The lack of significance of the relation between peer achievement and timing of first intercourse for girls may be related to gender differences reported in previous studies. Specifically, girls are found to develop larger social networks than those formed by boys (Coates, 1987; Pollard, 1993). Perhaps girls are more likely than boys to have peers and other support providers beyond the students found in their academic circle determined via tracking. Such an extended network would decrease the importance of the peer group acquired through school tracking programs.

Despite the contributions of this study, four precautions must be brought to the readers’ attention. First, the data used in this article were not collected with the express purpose of predicting sexual experience, thus the measures were not exclusively tailored for this article. Although we argue for the inclusion of certain measures in our model, this does not mean that these measures are optimal.

The second issue is related to the timing of first intercourse measure. Retrospective reports of age of first intercourse were used. We relied on young adults’ reports of first intercourse, assuming that their memories were accurate and truthful. Also, some respondents were already sexually experienced prior to age 13 (1% of girls and 4% of boys).[2] Therefore, some persons had sex prior to when the predictors were measured. Future studies should attempt to describe biological, psychological, and social characteristics of individuals and their sexual behavior prospectively, and then follow these individuals across the adolescent and young adult years. By doing this, we could begin to describe the development of sexual behavior across time and examine potential variability of influences across the life span. Predictors may differ in influence across adolescence, as individuals develop into young adults. Additionally, asking persons directly about their reasons for participating in sexual activity may shed some light on potential sources of influence.

Third, the study sample represented less than 40% of the overall MSALT sample. The attrition sample and the study sample differed considerably. Four of the nine predictors used in the models varied significantly between the samples. Two of these four measures, association with high-achieving peers and dating alone, were significant predictors of first intercourse. On this count, great care must be applied to the generalizability of these findings. Although the study sample appears to have had an edge in delaying intercourse compared to the larger attrition sample, no mean differences were found for the age of first intercourse when comparing the study sample to the attrition sample. Limiting the sample to White adolescents also reduces the generalizability of the study. The final precaution relates to the fact that the early initiating group may include individuals who experienced early coercive sexual situations. According to the AGI (1994), 74% of women who had sex before the age of 14 did so due to coercion. In addition, in the National Health and Social Life Survey (Laumann, Gagnon, Michael, & Michaels, 1994), about 25% of the female respondents reported they had not wanted their first sexual experience to occur. Despite our focus on “voluntary sexual intercourse,” some participants in this study may have had coercive experiences their first time. The early initiators are the most likely to have had a sexually coercive experience prior to their first voluntary intercourse (Miller, Monson, & Norton, 1995). In this sample, 4.4% of the girls and 5.3% of the boys reported first intercourse before age 14.

In spite of these precautions, this study provides a great deal of information regarding the predictors of timing of first intercourse. Sexual behavior can be influenced by biological, social, and psychological factors, depending on the model of focus. This multidimensional set of factors has implications for program and policy intervention, particularly given that the predictors are primarily from data collected when the respondents were in the seventh grade. The significance of these early predictors encourages creating developmentally appropriate prevention programs that target early adolescents (Bogenschneider, 1996). Indeed, the majority of adolescent sexuality programs available at the national level target early adolescents (Meschke, Bartholomae, & Zentall, 2000).

Based on the significant predictors that emerged from this study, potential programs to delay the timing of first intercourse need not be overtly associated with sexual behaviors. The final model in which all three types of predictors were considered supports intervention efforts in both the psychological and the social area of the adolescent’s life. For example, early adolescent prevention programs that serve to delay dating alone for girls or to enhance the expansion of social networks of boys may delay the timing of first intercourse. Attempts to intervene on the significant predictors would be an excellent test of the models put forth in this article (Bronfenbrenner, 1979).

Differences in the patterns of significant predictors by gender indicate that programs targeting particular issues by gender in early adolescence may be more effective. The normative nature of adolescent sexual behavior, current rates of adolescent pregnancy, and contraction of STIs, including HIV, warrant that the factors of adolescent sexuality be further examined. Currently, very few adolescent sexuality programs are tailored specifically for boys or girls (Meschke, et al., 2000).

More basic research is also needed regarding the prediction of the timing of first intercourse. The relation between additional biological and social predictors of first intercourse requires further examination, including the interactional effects. In addition to better understanding the precursors associated with early intercourse, it is equally beneficial to further examine those individuals who are sexually inexperienced as young adults. In general, the latter group has received very little attention from researchers.

In sum, each of the models, except the biological model for the boys, were predictive of the timing of first intercourse. It was hypothesized that the biopsychosocial model would prove to be the most effective model. The biopsychosocial model was a stronger model than the biological models, the girls’ biopsychological model, and the boys’ biosocial model. Although no significant increase in the biopsychosocial model’s prediction was found when compared to the girls’ biosocial model and the boys’ biopsychological model, a psychological predictor (importance of popularity) did remain significant in the biopsychosocial model for the girls and a social factor (peer achievement) remained significant in the boys’ biopsychosocial model. Thus, the importance of incorporating measures from the biological, psychological, and social arenas in predicting the timing of first intercourse has been demonstrated.

Adolescent Romantic Behaviors and Perceptions: With Family and Peer Relationships

This study examined the role of age, gender, and dating experience in adolescent romantic behaviors and perceptions. In addition, the linkage between the quality of relationships with parents and peers, and affective intensity with a romantic partner was investigated. Interviews were held with 168 Israeli adolescents, who also completed questionnaires assessing romantic behaviors, romantic perceptions, and quality of relationships with their romantic partner, parents, and a same-gender close friend. Older adolescents were more likely than younger adolescents to have a romantic partner. Romantic perceptions pertaining to companionship and excitement were less frequent among older adolescents. However, aspects such as intimacy and level of emotional involvement were similar across adolescence. Girls emphasized more attachment and care in their romantic relationships than boys. In addition, adolescents who were currently dating perceived romantic relationships more in terms of emotional involvement than adolescents who were not dating. The affective intensity with same-gender close friend but not with parents was linked to the affective intensity with the romantic partner.

Although adolescent romantic experiences are believed to play an important role in the development of later significant romantic relationships, little is known about their nature. The literature on adolescent romantic relationships focuses on patterns of dating and sexual behavior. Early relationships with romantic partners are casual, less intense, and short-lived (Feiring, 1996). Young adolescents focus more on who they are, how attractive they are, how they should interact with the other gender, and how this interaction will be accepted by their peers (Brown, 1997). Gradually, interest and interaction with the other gender increases, and at 15 years of age the majority of adolescents have had some dating experience (Blyth, Hill, & Thiel, 1982). Sexual activity takes place in dating, and adolescents perceive having some sexual experience by the age of 15 as normative (Feldman, Turner, & Araujo, 1999). Conceptually, relationships with the other gender were more understood within the framework of friendship or companionship (Blyth & Foster-Clark, 1987; Feiring & Lewis, 1991). The classic work of Dunphy (1963) described a gradual change in the nature of cross-gender interactions. This starts with a preference to be in places where other gender peers are likely to be present, then advances to participation in mixed-gender group activities, group dating in which couples meet in the context of the “crowd,” and individual dyads going out on their own.

Other studies investigated adolescent romantic relationships within the framework of the network of relationships with significant others such as parents, same-gender friends, and siblings (Blyth et al., 1982; Connolly & Johnson, 1996; Furman & Buhrmester, 1992; Lempers & Clark-Lempers, 1993). In these studies, adolescents were asked to rate the support that they received in a particular relationship, and its functional importance, or to indicate the range of activities that they engaged in with the various figures (Laursen & Williams, 1997). Results consistently showed that with age the importance of romantic relationships as providers of support and targets of intimacy increases, although friends and family members remain important figures in the life of teenagers throughout adolescence.

Studies that focus on the romantic experience and its qualitative nature have been conducted mainly among college students or adults. Sternberg’s (1986) approach focused on the functions of romantic relations, and outlined three major components–intimacy, passion, and commitment. Hazan and Shaver (1987) conceptualized love relationships in terms of attachment. According to this view, a romantic partner is thought to become an attachment figure, replacing the parental attachment figures as uppermost in the attachment hierarchy. Hazan and Shaver described romantic love as an adaptive biological process, which facilitates attachment between sexual partners. The attachment between romantic partners differs to some extent from the parent-child bond as it involves the integration of attachment, caregiving, and sexual behavior between two adults.

This study was designed to address age-related differences in adolescent romantic experience and perceptions, especially identifying the qualitative differences in the nature of adolescent romance during the different stages of adolescence. Furman and Wehner’s (1994) theory on adolescent romantic relationships is a pioneering attempt to explain how cross-gender encounters initiated in a group context develop and change into attachments. Based on their integration of the theoretical works of Bowlby and Sullivan, Furman and Wehner suggested that younger adolescent romantic partners serve more as companions and friends, thus providing experiences of cooperation and reciprocity. In addition, these relationships are also based on self-disclosure and consensual validation of self-worth. At a later stage, a partner is sought out in times of distress, and is expected to provide support, comfort, and caregiving. In conjunction with the emergent sexual interest the individual turns to the romantic partner for sexual fulfillment as well. By fulfilling the different needs, the romantic partner becomes part of the hierarchy of important figures and its importance for the adolescent in comparison with other important figures moves up with age (Furman & Buhrmester, 1992).

In a more recent article, Furman and Wehner (1997) elaborated on the evolution of a romantic partner into an attachment figure, placing extra emphasis on the importance of the interaction experience for the quality of a romantic relationship: “As individuals grow older and acquire more experience in various romantic relationships, they may be more likely to turn to a partner to fulfill these functions than when they were younger and less experienced” (p. 23). Studies in social exchange theory suggest that mutually beneficial exchanges promote future cooperation as participants learn to rely on one another for rewards (Kelly et al., 1983). Thus, gaining more experience with romantic relationships, in addition to chronological age, may also contribute to changes in romantic behaviors and in how romantic love is understood.

In line with these conceptualizations, clear changes with age can be expected. Affiliative behaviors and motives can be found among romantic partners during the earlier stages of adolescence, reflecting the companionship quality of the interaction with members of the other gender. The caregiving and attachment components are expected to become more significant in romantic relationships in later stages of adolescence and in young adulthood, when the need for a new attachment figure increases. In addition, experience and acquisition of some competencies in interacting with the other gender may also facilitate the coordination of affiliative, attachment, and sexual needs, and their becoming central to the relationship (Furman & Wehner, 1997).

Evidence to date suggests that there are gender differences in friendship intimacy. Adolescent female friends are reportedly closer, and more inclined to self-disclosure, than are boys (Camerena, Sargiani, & Petersen, 1990; Jones & Dembo, 1989; Sharabany, Gershoni, & Hofman, 1981; Shulman, Laursen, Kalman, & Karpovsky, 1997). However, in adolescent and young adult cross-gender friendships (Reisman, 1990), adolescent romantic partnerships (Shulman, Levy-Shift, Kedem, & Alon, 1997), and married couples (Merves-Okin, Amidon, & Berndt, 1991) no significant differences were found between boys and girls on self-disclosure and expression of emotions. Intimacy, measured in terms of commitment and communication, also showed no significant differences within couples (White, Speisman, Jackson, Bartis, & Costos, 1986).

Nevertheless, girls spend more time with and thinking about boys (Richards, Crowe, Larson, & Swarr, 1998), and more perceive their romantic relationships as supportive (Connolly & Johnson, 1996). Male adolescents reported a lower level of respect for their romantic partner than did female adolescents (Shulman et al., 1997), and men are more likely than women to interrupt their partners (Leaper & Anderson, 1997). Girls are thus more attentive to their partners and more perceive relationships as supportive. This is in line with Gilligan’s (1982) contention that women are more at ease being close in their relationships and their life is organized around being able to maintain affiliations. In addition, women were traditionally expected to stay connected in meaningful ways and care about their partners and families (Papp, 1989). Previous research shows that women report more frequent and positive caregiving behaviors than men, probably reflecting societal expectations that reward female nurturant behavior (Kunce & Shaver, 1994). Overall, we may expect that female adolescents will emphasize more commitment and care in their romantic relationships than male adolescents.

The growing interest and investment in a romantic partner cannot be detached from the two other central relationships in an adolescent’s life, namely family and close friends. It is well documented that throughout adolescence frequency of interactions and diversity of activities shift from parents to peers (Laursen & Williams, 1997). Nevertheless, one cannot overlook the fact that parents remain a major source of support for their adolescent children even as the salience of peers increases (Youniss & Smollar, 1985), and often maintain this status throughout the college years as well (Furman & Wehner, 1997). Moreover, understanding romantic relationships from a developmental perspective and within an attachment framework (Furman & Wehner, 1994) may suggest that the quality of the romantic relationship is related not only to the experiences with the romantic partner, but also to the quality of relationship with parents and peers. Thus, relational patterns such as the sense of closeness experienced in the relationship with parents and peers are linked to the sense of closeness experienced with a romantic partner. This assumption is in line with attachment theory. Models of self and others, and relationships that develop from experiences in close relationships, influence the nature of interaction with the environment, expectations concerning availability, responsiveness, and attitudes of others, as well as expectations about the self in future relationships (Sroufe & Fleeson, 1986).

In sum, three major questions are addressed in this study:

Do romantic behaviors and perceptions differ across adolescence? We hypothesized that older adolescents will be more likely to report having a romantic partner, will report on romantic relationships that last longer, will meet frequently with, and will be more likely to be sexually intimate with their romantic partner than younger adolescents. In addition, it was hypothesized that younger adolescents will be more likely to perceive romantic relationships in terms of friendships, whereas older adolescents will be more likely to perceive romantic relationships in terms of care and attachment. It was also expected that more experience in romantic relationships will be related to higher competence in interacting with the other gender and will be reflected in higher levels of involvement with and care for the romantic partner.
Do romantic behaviors and perceptions differ across gender? We expected that girls will be more likely to perceive romantic relationships in terms of caregiving and support, whereas boys will be more likely to perceive romantic relationships in terms of companionship.
To what extent do relationships with parents and close same-gender friends relate to the quality of the romantic relationship during adolescence? We explored the extent to which the positive dimension of relationships with parents and peers, namely the level of experienced closeness in these relationships, is related to the sense of closeness experienced in an adolescent’s romantic relationship.
METHOD
Participants

A total of 168 adolescents (91 boys and 77 girls) participated in the investigation. Participants were drawn from four high schools in medium-sized towns in the center and south of Israel, and represented a lower-middle-class to an upper-middle class spectrum. Participants were included after they gave consent. Ninth-grade participants were included in the study only after parental consent was given as well, whereas 11th and 13th graders were included in the study after they consented. When our study was conducted, personal consent for participating in psychological study was approved for 11th graders. The 168 adolescents represented a participation rate of 87%. All the participants lived with their families. The sample consisted of 54 ninth graders, 30 girls and 24 boys (mean age = 14.11 years, SD = 0.37); 53 eleventh graders, 30 girls and 23 boys (mean age = 16.26 years, SD = 0.52); and 61 thirteenth graders, 31 girls and 30 boys (mean age = 19.18 years, SD = 0.59). The 13th graders attended two senior high schools where studies continue for 4 years, instead of the more common 3 years. These schools are not academically or socially different from schools that offer 3 years of senior high school besides offering an additional year of schooling that mainly focuses on science education. The decision for an additional year of schooling is made by students themselves and not dictated by the school system. No academic, socioeconomic, or family background differences between those who continue to study an extra year and those who graduate from school after the 12th grade were reported by the schools’ principals. These 13th graders form a unique comparison group as they represented a group of older adolescents who are not away from home and who study in a high school setting and not a college.

Measures and Procedure

Participants were interviewed, and were asked to complete questionnaires assessing the nature of their romantic behaviors, their romantic perceptions, and their relationships with a romantic partner, parents, and a same-gender close friend. Participants were individually interviewed and completed questionnaires in groups of 20 students at their schools. Participants randomly were interviewed first or completed questionnaires first.

Romantic Behaviors

Participants were asked to indicate in open-ended responses the following information:

Occurrence of dating (currently dating, have dated, have never dated).
Length of current or most recent significant romantic relationship, in weeks.
Interaction frequency, in minutes, of contact with the romantic partner during a morning, an afternoon, and an evening in the preceding week. The total time of contacts represented the interaction frequency with the romantic partner on an ordinary day in the preceding week.
Participants were asked to rate on a scale ranging from 1 (never touched each other) to 9 (had sexual intercourse) how sexually intimate they were with their current romantic partner or with their previous partner.
These open-ended questions were patterned according to Berscheid, Snyder, and Omoto’s (1989) revised version of the Relationship Closeness Inventory. Berscheid and her colleagues found reports of frequency of interaction to possess acceptable test-retest reliability over a period of 3 to 5 weeks (.81).

In addition, to measure the extent of perceived closeness with the romantic partner, participants were asked to rate the affective intensity with their current or most recent romantic partner. The Index of Affective Relationships (Takahashi & Nagima, 1994) was used to measure intensity of affective relationships. In its original form this is a self-report instrument consisting of 18 statements describing affective desires to rely on or to be with a significant figure in the life of the adolescent. In this study, an adapted and shortened version of the instrument was used, comprising 13 items. For each item respondents were required to rate on a scale ranging from 1 (never correct) to 5 (always correct) the extent to which a statement characterized their relationship with a certain figure. Examples of statements are “When I receive bad news I want to be with X;” “I want to share pleasant feelings with X.” In this study, respondents were asked to complete the instrument with regard to their romantic partner as well as with regard to their mother, father, and same-gender close friend. Mean score for affective intensity with each figure was computed. Mean scores ranged from 1 to 5. Cronbach alphas for the four figures in this shortened version were satisfactory and ranged from .88 to .95.

Romantic Perceptions

1. Romantic relationship experiences were measured by Hendrick and Hendrick’s (1986) measure of love styles. The 36-item measure indexes six love styles. A respondent is asked to indicate the extent to which a certain statement characterizes the relationship with the romantic partner on a scale ranging from 1 (does not characterize the relationship at all) to 5 (very much characterizes the relationship). The six love styles, and examples of items, are listed next.

Eros (physical love): “We were attracted to each other immediately when we first met.”

Ludus (game-playing love): “I try to keep her uncertain about my commitment to her.”

Storge (friendship, companionate love): “The best kind of love grows out of a long friendship.”

Pragma (practical, objective love): “An important factor in choosing a boyfriend is how he will be as a father.”

Mania (preoccupation with the romantic relationship): “When I am in love, I have trouble concentrating on anything else.”

Agape (caregiving, self-sacrificing love): “I try to always help her through difficult times.”

For each love style a mean score was computed, hence scores ranged from 1 to .5. Cronbach alphas for the six styles in this study were as follows: Eros, alpha = .76; Ludus, alpha = .71; Storge, alpha = .73; Pragma, alpha = 65; Mania, alpha = .74; Agape, alpha = .76. Removal of one item from the Pragma scale resulted in an alpha of .69. The mean for the Pragma love style was thus computed based on five items. Intercorrelations among love styles revealed the Eros, Mania, and Agape love styles to be interrelated (correlations ranged from .43 to .67). Therefore these three styles were collapsed into one love style reflecting attraction, involvement, and care, and was termed Emotional Involvement love style.

2. Perceived advantages of having a romantic partner were examined by Feiring’s (1996) semistructured interview about dating in which adolescents are requested to talk about the advantages of having a boyfriend or girlfriend. Content analysis was conducted on adolescents’ transcribed comments to determine categories that are most frequent and best describe the data. The following advantage categories were derived:

Companionship: “You have somebody to go out with, to study together with before exams.”

Intimacy: “To share my personal matters with someone.”

Attachment: “I feel that somebody, who is not a family member, really loves me, and I can count on him.”

Care: “Somebody who spoils you and you spoil her.”

Status: “In my class, it makes me a MAN.”

Sex: “You can kiss him and do things you do not do with a friend.”

Excitement: “It is just great to think that you are involved and to wait for the next time you meet.”

Each participant’s answers were coded according to whether they mentioned or did not mention a given category. Categories are not mutually exclusive, and a participant could mention a number of advantages. To establish interrater reliability on these categories, transcripts were coded independently by two raters. Cohen’s Kappas were in the range of .87 to .91. The interview data used in this study are thus highly reliable and provide adolescents’ perceptions of their romantic relationships. The categories derived in this study are quite similar to those derived by Feiring. In sum, responses fell into one of two possibilities: A respondent has mentioned or has not mentioned a category.

RESULTS
Plan of Analysis

In this section, we first describe age, gender, and dating experience differences in romantic behaviors and perceptions. Then we examine the linkages between affective intensity with parents and peers, and affective intensity with a romantic partner during adolescence.

Age, Gender, and Dating Status Differences in Romantic Behaviors

Frequencies and percentages of three categories of dating status (currently dating, have dated, have never dated) were tabulated across three age groups and are presented in Table 1. Sixteen and 19-year-olds reported a higher frequency of dating currently than 14-year-olds. In addition, 19-year-olds reported the highest frequency of having never dated. However, 14-year-olds reported the highest frequently of having dated in the past, whereas 16-year-olds in comparison with them reported less that they had dated in the past, X2(4,N = 168) = 28.70, p < .0001.

To assess age, gender, and dating status differences in romantic behaviors the following analyses were conducted on the cohort of adolescents who reported currently dating and those who had dated in the recent past. A 3 (age) x 2 (gender) x 2 (dating status: currently dating and have dated in the past) analysis of variance (ANOVA) was conducted to assess the mean duration of romantic relationships across three age groups and gender among those who were currently dating and those who had dated in the recent past. Results revealed two significant main effects: for age, F(2,114) = 7.05, p < .001; and for gender, F(1,114) = 3.54 p < .05. A Scheffe follow-up contrast (p <.05) revealed that 14-year-olds compared with 16- and 19-year-olds were engaged in relationships that last longer, M = 52.29 weeks (SD = 40.48) versus M = 26.88 weeks (SD = 42.35) and M = 20.81 weeks (SD = 24.02), respectively. Girls in comparison with boys reported engaging in relationships that lasted longer, M = 38.57 weeks (SD = 39.62) versus M = 28 weeks and M = 27.90 weeks (SD = 37.32), respectively.

A 3 (age) x 2 (gender) x 2 (dating status) ANOVA was conducted to assess the mean length of time, in minutes, during which adolescents were engaged with their romantic partners in the preceding week (or were engaged in their most recent romantic relationship). Results revealed one significant main effect for dating status, F(1,114) = 5.74, p < .02. Those who were currently dating reported spending more time with their romantic partners than adolescents reporting on time spent in the past with a romantic partner in a relationship that had terminated, M = 152.57 min (SD = 109.95) versus M = 86.92 min (SD = 173.61), respectively.

An additional 3 (age) x 2 (gender) x 2 (dating status) ANOVA was conducted to assess the level of sexual intimacy with the romantic partner. Results revealed one significant main effect for dating status, F(1,114) = 10.09, p < .01. Those who were currently dating reported being more sexually intimate with their partner than those who reported on the sexual intimacy in a relationship that had ended, M = 7.44 (SD = 2.08) versus M = 5.69 (SD = 2.27), respectively.

Finally, a 3 (age) x 2 (gender) x 2 (dating status) ANOVA was conducted to assess the mean level of affective intensity with the romantic partner across three age groups and gender among those who were currently dating and those who had dated in the recent past. Results revealed two significant main effects: for gender, F(1,114) = 5.28, p < .05; and for dating status, F(1,114) = 4.78, p < .05. Girls reported a higher level of affective intensity in their romantic relationships than boys, M = 4.42 (SD = 0.46) and M = 4.10 (SD = 0.74), respectively. Adolescents who were currently dating reported higher affective intensity with their partner than adolescents reporting on a relationship that had ended, M = 4.46 (SD = 0.53) and M = 4.10 (SD = 0.66), respectively.

To summarize, among older adolescents there was a higher chance that they were currently engaged in a romantic relationship; however, 14-year-olds reported more than 16-year-olds that they had dated in the past. Romantic relationships of 19-year-olds lasted longer than those of younger adolescents. Girls across all age groups were engaged in relationships of longer duration and reported a higher level of affective intensity with their romantic partner. Finally, currently dating adolescents reported a higher frequency of interaction and higher levels of sexual intimacy and affective intensity in their relationships than adolescents reporting on relationships that had ended.

Age, Gender, and Dating Status Differences in Romantic Perceptions

The following analyses were conducted on the whole sample. The three dating status categories, currently dating, have dated, and have never dated, represent the independent variable of dating experience.

Love styles. A 3 (age) x 2 (gender) x 3 (dating status: currently dating, have dated, and have never dated) multivariate analysis of variance was conducted to assess the mean level of the love styles across age, gender, and dating status. Three main effects emerged: for age, F(2,159) = 2.79, p < .01; for gender, F(1,159) = 3.59, p < .001; and for dating status, F(2,159) = 3.00, p < .001. No significant interactions among age, gender, or dating status were found. Follow-up ANOVAs and Scheffe contrasts (p < .05) specified age, gender, and dating status differences on each love style.

Univariate ANOVAs revealed a significant age difference for the Storge love style, F(2,159) = 4.81, p < .001. A Scheffe follow-up contrast showed that perception of romantic love as friendship was higher among 14- and 16-year-olds, M = 3.74 (SD = 0.55) and M = 3.65 (SD = 0.75), than among 19-year-olds, M = 3.41 (SD = 0.68). An additional set of univariate ANOVAs revealed a significant gender difference on the Ludus (game-playing love) style, F(1,159) = 11.03, p < .001. Boys across the three age groups described a higher level of Ludus-style love than did girls, M = 2.86 (SD = 0.71) and M = 2.39 (SD = 0.62), respectively.

A third set of univariate ANOVAs revealed a significant difference in the Emotional Involvement love style, F(2,159) = 9.13, p < .001. A Scheffe follow-up contrast showed that adolescents who were currently dating reported a higher level of the Emotional Involvement love style than adolescents who had dated, M = 3.85 (SD = 0.63), and M = 3.36 (SD = 0.67), respectively. Adolescents who had never dated reported an intermediate level of the Emotional Involvement love style, M = 3.59 (SD = 0.43).

Advantages of having a romantic partner. A series of 3 (age) x 2 (gender) x 3 (dating status: currently dating, have dated, and have never dated) x 2 (advantage category: was mentioned or was not mentioned) log-linear analyses were conducted. Results revealed several significant two-way interactions between independent variables and advantage categories. Two log-linear analyses revealed interactions between age and the companionship advantage category, X2(2,N = 168) = 6.65, p < .05, as well as between age and the excitement advantage category, X2(2,N = 168) = 10.26, p < .01. As can be seen in Table 2, 14- and 16-year-olds mentioned more companionship as an advantage of romantic relationship than 19-year-olds. In addition, 19-year-olds mentioned excitement as an advantage of romantic relationships less than did younger adolescents.

In addition, two log-linear analyses revealed interactions between gender and the attachment advantage category, X2(1,N = 168) = 30.36, p < .0001, as well as between gender and the care advantage category, X2(1,N = 168) = 33.84, p < .0001. Attachment and care were more perceived by girls across all age groups and dating statuses as an advantage of having a romantic partner than by boys (See Table 2). An additional log-linear analysis revealed a significant interaction between dating status and sex as an advantage category, X2(2,N = 168) = 7.88, p < .01. Adolescents who were currently dating or had dated in the past were more likely to mention sex as an advantage of romantic relationships than were adolescents who have never dated (see Table 2).

To summarize, older adolescents (19-year-olds) perceived romantic relationships less in terms of friendship, and considered companionship less to be an advantage of having a romantic partner, than did younger adolescents (14- and 16-year-olds). In addition, 19-year-olds mentioned excitement as an advantage less than did younger adolescents. No age differences were found in perceptions of romantic relationships in terms of intimacy, caregiving, or attachment. Attachment and care were more perceived by girls across all age groups as an advantage of having a romantic partner than by boys. In contrast, the perception of romantic relationships in terms of game-playing love was more emphasized by boys than by girls. Adolescents who were currently dating perceived romantic relationships more in terms of emotional involvement than did adolescents who currently did not have a romantic partner. In addition, sex was more frequently mentioned as an advantage for dating among adolescents who were currently involved in a romantic relationship or had dated in the past than it was among adolescents who had never dated.

Intensity of the Affective Relationship With a Romantic Partner Versus the Role of the Quality of Relationships With Parents and Peers

A multiple regression was conducted to determine the linkages between the quality of relationships with parents and peers and the intensity of the affective relationship with a romantic partner. (This analysis was conducted with 120 adolescents who reported currently having a romantic partner and who had a partner in the past.) Results of the multiple regression are presented in Table 3. As can be seen, the affective intensity with parents did not explain affective intensity with a romantic partner, whereas closeness to a same-gender friend explained 43% of affective intensity with a romantic partner.

To control the possible contribution of other variables to the affective intensity with the romantic partner, a hierarchical regression was performed in which demographic variables (gender and age) as well as variables related to the romantic experience and perceptions (the status of dating, the frequency of contact with the romantic partner, the sexual intimacy with the romantic partner, and romantic perceptions in terms of love styles) were inserted prior to the insertion of the indexes of affective intensity with mother, father, and close same-gender friend. Results showed that even after controlling those variables, affective intensity with a same-gender close friend explained 27% of the sense of closeness with the romantic partner, whereas affective intensity with either of the parents did not explain affective intensity with a romantic partner.

DISCUSSION
Results of this study show that almost one half of the 16- and 19-year-olds reported being currently engaged in a romantic relationship. Among 14-year-olds, only 15% reported being currently involved in a romantic relationship. These findings extend results from previous studies reporting that the proportion of individuals who state that they have a boyfriend or a girlfriend increases with age (Furman & Buhrmester, 1992), and that by the age of 15 most teenagers have had some experience of dating (Blyth et al., 1982; Feiring, 1996). Report of previous dating, however, showed a different age-related trend. The majority of 14-year-olds (65%) reported having had some previous experience with dating, whereas only 22% of 16-year-olds reported this. This unexpected trend may indicate a previously unnoticed aspect in the study of adolescent romance.

In the majority of studies, adolescents are asked to indicate whether they are currently dating, have dated in the past, or have never dated. However, to the best of our knowledge, no criteria were ever formulated to define the terms boyfriend or girlfriend. Previous studies showed that adolescent romantic relationships are part of the adolescent social context and experience (Connolly & Johnson, 1996; Dunphy, 1963). Our results showed that conceptions of companionship and friendship are an integral part of adolescents’ understanding of romance. It is reasonable to question whether adolescents, especially younger adolescents, differentiate between a romantic relationship and a cross-gender friendship. We know that children and adolescents are involved in cross-gender interactions or friendships (Leaper & Anderson, 1997), but we do not know when and how a friendship tums into a romantic relationship or is considered as such. In a recent article, LaVoie, Johnson, Mahoney, Ramet, and Anderson (1998) described a continuum of dating status, starting from “close friends” via “casual dating” to “exclusive dating.” We speculate that 14-year-olds may regard a close cross-gender friendship or casual dating as an exclusive romantic relationship. Going out in mixed-gender groups at this age might be considered dating and this might be the reason why many 14-year-olds compared to 16-year-olds tend to report they have experience in dating. Sixteen-year-olds probably have a stricter idea of what it means to have a boyfriend or a girlfriend or to be in love. A recently published study gives rise to this possibility. Montgomery and Sorrel (1998) reported a trend suggesting a decrease in the number of adolescents reporting on having been in love among 11th graders.

This possible changing dialectic of friendship and romance is further evinced in the age-related differences in romantic perceptions found in this study. Some perceptions of romantic relationships are similar across adolescence, although others differ. Experience of romance as a caregiving relationship, as well as the value placed on intimacy and attachment is predominant across the various stages of adolescence. In these aspects adolescent romantic relationships, even among younger adolescents, resemble those of adults (Levesque, 1993). Yet despite these consistencies, there are aspects that differ with age. In this study, younger adolescents emphasized aspects of friendship or companionship in their romantic relationships, whereas this tendency was lower only among the 19-year-olds. In addition, there was a decrease in the perception of excitement related to the romantic relationship. The age differences in perception of a romantic partner as a companion are in line with Furman and Wehner’s (1994) theory that at an earlier age romantic partners serve more as companions and friends, providing experiences of cooperation and reciprocity. At a later stage when a partner is more expected to provide support, comfort, and caregiving, and turning into more of an attachment figure, the aspects of companionship and friendship in the romantic relationship are less pronounced.

The age-related decrease in excitement points to an additional aspect of romance in adolescence. Phenomenological studies, as well as lay intuition, have described components such as uniqueness, absolutes, and idealism as characterizing adolescent romance (Fischer & Alapack, 1987). Fascination has been used to differentiate romantic from friendship relations (Hatfield & Rapson, 1987). However, as Feiring (1996) suggested, fascination “captures the brief but intense nature of adolescent romance. Given the amount of attention focused on the dating partner, short-term fascination might best describe the initial stage of romance in adolescence” (p. 192).

Whereas age is related to a decrease in characterization of romantic relationships in terms of companionship and fascination, current experience of dating was found to be related to an increase in aspects characterizing closeness and mutual commitment. This can be understood because the romantic relationship provides rich opportunities for cooperation, mutualism, and reciprocal altruism (Laursen & Williams, 1997). As relationships develop, individuals may become more confident in partners’ availability and support, and this further affects romantic perceptions. Furman and Wehner (1997) found that college women in exclusive relationships were more secure and less preoccupied in their romantic styles than those dating more casually. Kirkpatrick and Hazan (1994) found that some individuals with insecure avoidant styles switched to a secure style when they become involved in a new relationship. However, some caution is warranted. It is possible that adolescents reporting on a relationship that had ended describe a less successful relationship, or tend to reconstruct a previous relationship in a less favorable manner.

Despite the importance of romantic experience, age differences in romantic behaviors and perceptions were found independent of dating experience. It is possible that age and dating experience are related to different aspects of adolescent romantic relationships. Experience is probably more related to the quality of the relationship whereas age-related trends are possibly bound to the meaning of love and its functions during the different stages of adolescence. However, this finding must be interpreted in light of the following limitation. Some cells, like 19-year-olds who had never dated or 14-year-olds who were currently dating, contained few participants, and in fact may raise questions whether interaction effects between age and experience exist but were not detected in this study. This must be tested in future studies.

Consistent gender differences across the three age groups emerged with regard to romantic behaviors and perceptions. Girls reported a higher level of affective intensity with their romantic partners. In addition, girls more than boys reported valuing attachment and care in their relationships. Boys, by contrast, perceived romantic relationships more as game-playing love. These findings recall the well-documented gender differences in adolescent close relationships. Adolescent female friends are reportedly closer, and more inclined to self-disclosure, than are beys (Camerena et al., 1990; Jones & Dembo, 1989; Shulman, Levy-Shift, Kedem, & Alon, 1997). Boys tend to express themselves through separateness, characterizing friendship in terms of shared activities, whereas girls perceive relatedness, emphasizing mutual closeness and reciprocity in friendship (Smollar & Youniss, 1982). Similarly, this finding recalls the higher tendency of women to report higher level of commitment, caregiving, and security than men (Gilligan, 1982; Kunce & Shaver, 1994; Pines, 1998).

The affective intensity with the romantic partner was related to the quality of the relationship with a same-gender close friend, whereas the relationship with parents was not related to affective intensity with a romantic partner. Our results consistently demonstrate that although adolescent romantic relationships contain components found in adult love experiences, components of friendship and companionship are also dominant in romantic relationships during earlier stages of adolescence. The friendship quality of adolescent romantic relationships can also be attributed to the fact that many of the interactions with the romantic partner take place within the peer arena. Adolescents meet in mixed-gender group activities, and date in a group context (Blyth & Padgham, 1987; Dunphy, 1963). Furman and Wehner (1993) also found that romantic relational styles of high school girls were related to relational styles with friends but not with parents. Among college women, however, romantic relational styles were related to those with friends as well as to relational styles with parents. Future studies on late adolescent and young adult romantic relationships could illuminate to what extent friendships moderate relationships with romantic partners, and if and when the role of relationships with parents becomes more prominent in explaining late adolescent and early adult romantic relationships (as found by Owens et al., 1995).

Some additional limitations of this study should be addressed. Our resuits demonstrated differences as a function of age, yet the study was cross-sectional in nature. A longitudinal study is needed to rule out alternative explanations. In addition, this study dealt with the reported behaviors and perceptions of adolescents with regard to romantic relationship in which they are or were involved. A relationship consists of two partners. As we have shown (Shulman et al., 1997), more can be learned about a relationship when both partners are examined rather than one partner alone.

In closing, our results show that adolescent romantic relationships consist of components found in adult romantic relationships, such as the mutual caregiving quality, but they are also an arena for joint activities and companionship, as is friendship. However, the nature and quality of adolescent romance was found to be different in early and late adolescence. In line with Furman and Wehner’s (1997) suggestions, future longitudinal studies of the development of adolescent romantic relationships could help us better understand the transformations in affiliative, attachment, caregiving, and sexual features.

A look at one relationship

Relationships are hard work. And you certainly don’t have to have autism for this to be true.

Thomas
I met Theresa in, of all places to meet someone, an America Online chat room. However, this particular chat room involved parents supporting other parents of children with autism. There was the occasional bad apple who would pop in every now and again, but for the most part it was very helpful to those who attended. Theresa lived in Virginia, and one day her work brought her to Ohio, where we met. As it was my 34th birthday, she brought me a bear that I have been sleeping with ever since.

Theresa is a challenge, with three kids from a previous marriage (two with autism; one diagnosed before we met and one after), and I am sure I am a challenge with autism. I am not sure I care to know which of us is more of a challenge. I don’t think it really matters.

I had been in relationships before. Two of them were serious. They didn’t work out, and I was at the point in my life where I had just about given up. I had seen autistic friends of mine get married and get divorced. I didn’t want that to happen to me. Was I really ready to try again? It was try again or be sad and lonely all my life. I knew I didn’t want to be sad and lonely, so I decided to try again. Problems on both sides immediately appeared. Her insecurities and my sensory problems made the relationship difficult.

Theresa
When I first met Thomas, I was just coming out of a divorce. It was a difficult marriage; having a child with autism makes any marriage more challenging. I had heard of Thomas McKean and his writing before I actually met him online. It was a time of searching for me because my youngest son was showing some strong autistic traits as well. I was primarily searching for answers, but what I found was a good friend. Thomas was compassionate and understanding of everything I was going through. He was the person I would call when that IEP did not quite go right. He was the person I could cry on when things were getting out of control.

We decided to make a go of it. My trust issues were a big concern to him, and his sensory issues were a huge concern for me. We started talking more on the phone. There were many nights when we would talk until 4:00 a.m. Thomas always had a way to make me laugh. Thomas’ charm, wit, and continued compassion were some of the reasons I fell in love with him. We lived about 7 hours away, and we did get to visit often. I remember the first time I went to visit. I met with Thomas on his birthday. I think we both knew how special our relationship would be then, but our insecurities held us back. It was several months later before we said we would try to date.

Thomas
Theresa seemed different from the others. She had an understanding about her that I found very hard to resist. She not only tried but was able to relate to me in a way no one had before. How could I not love her?

Theresa
We have now been dating for several months. Our commitment and love for one another is strong; however, there have been continued problems. Many couples have little disagreements and concerns, especially when first dating. Ours seemed to be amplified by problems associated with autism. I felt like, “This is great. Now I have autism times three in my life.”

Thomas is moving to Bedford to be with me and the kids, and I have supported him in that decision. He has had the desire to leave Ohio for quite some time, and my little town seems to be ideal for him. The fact that I am here helps, too. I am excited about his move, and I have made the commitment, as he has, to make this relationship work. There are some days I have doubts, though, as I am sure we all do about people we get involved with. I am also very much aware that the move will be very difficult for Thomas–a transition that will pose its problems to everyone. I am proud of him for this decision, and it shows me how much he is willing to make it work for us. Ordinary people would be in the boat that says, “It’s not a big deal for me to move to the people I love,” but I know that for Thomas it is a strong challenge to move to be with us.

Thomas
It is true that I have wanted to leave Ohio for quite some time. Aside from a few close friends, there is nothing keeping me there. Columbus is not a very pretty place, nor is it very friendly. Having done a lot of traveling the past several years, I have seen that there are places far, far better to live than Columbus.

Bedford is a wonderful place. When you walk outside in Columbus, you see the freeway. When you walk outside in Bedford, you see mountains. When you walk down the street in Columbus, people go out of their way to avoid you. When you walk down the street in Bedford, people stop and say hello. No one is moving at warp speed in Bedford like they are in Columbus. The saddest part of people moving so fast in the big cities is that while they are moving fast, they really don’t know where they are going.

I have made several friends here in Bedford already. It is frightening to leave a place you have called home for 34 years, but it is also exciting to be starting over.

Theresa
There are many problems we face as a couple that “typical” relationships do not encounter. The biggest problem in our relationship comes from sensory issues. The second time I met with Thomas, I kissed him on the neck to say “Hi” to him. He pushed me away and said it hurt. During that trip, I was also a little sweaty from the long drive, and I got pushed away and told to take a shower and then he would hug me. Warning bells went off. I was thinking to myself, “I just drove 5 hours to see you, and you are pushing me away for kissing you.” Well, I am finding out that is only the tip of the iceberg. The tactile issues are so severe at times that intimacy seems like an impossible task some days. I need to always remember not to take it personally that Thomas needs to be touched in a certain manner so as not to cause him to be in pain. I believe we all have ways that we prefer to be touched over others. Thomas has to have deep pressure. I cannot go up to him and caress his arm or hand. I have to go give him a bear hug or rub his arm, rather than the simple caress that I prefer. Touch on certain areas of his body is a cause for pain. On his neck and face, touch is enough to cause him to cringe, as it physically hurts him. Thomas has been great about telling me what he does prefer and what brings him pleasure. I was very hurt by how roughly he would push me away, so although he does still cringe, now he says it tickles in a more gentle tone. He shows me what touch is pleasing to him and how to do it. I, on the other hand, show him that I don’t like a strong touch, but I prefer a light, gentle one. I also know that this is difficult for him at times, but he gets a 10 for effort. With good communication, this is an area in our life that has improved a great deal. I must say this about Thomas–and I cannot say it enough–he is always willing to try. His effort to make our relationship work has been worth it all, even on those tough days.

Thomas
One of the things I find most difficult in our relationship is Theresa’s occasional tendency to get upset about being on time. A good example of this is that sometimes in the morning she will tell us all to hurry up if we have to be somewhere. I have to keep reminding myself that it is nothing personal and that she just wants to be on time. This can be hard for me because I have a way of taking these things personally.

Her kids can be a challenge for me, too. I am in the very awkward position of not knowing how far to go or not to go to have fun or to discipline. Theresa changes her mind about this constantly. I am sure that as the relationship goes on, it will settle into a routine regarding those issues, but right now I am not sure what to do. A lot of times I come to her for help with the kids. Sometimes she tells me it was the right thing to do, and sometimes she gets mad at me that I didn’t handle whatever the problem was myself. This is something I find to be very confusing.

Theresa
Another hurdle that we continue to try to tackle is our differences in motivation styles. I get up at 5:30 a.m. and go all day, until I pass out again around 11:30 p.m. With three children, being a single parent, taking graduate courses, and working part time as well as on all the boards and volunteer work, I am up all day and running. I enjoy this; I like to keep busy. Thomas will go to bed leisurely around 2:00 a.m. and get up around 1:00 p.m. or so the next day. We discussed this for quite some time because when Thomas comes to visit me, I want him to join me in some of my activities. So we finally agreed on a solution there as well. Thomas said that he had “balls” he could juggle in a day. He explained what could cause him “to lose a ball,” such as going to a meeting with me, but that by playing frisbee with JR maybe he could gain one back. He would communicate when too many balls were being juggled and he was overstimulated. I know, after being with Thomas for so long now, when he feels it’s too much; he clings to me and constantly wants deep pressure, especially through being hugged. I look for these external clues as well as for his verbal ones. I am learning to join him in his circle and find out what his needs are. When I manage to do this, Thomas is much more relaxed, and therefore I am as well.

Thomas
The same can be said of me. When she is happy, I am happy. I love to see her smile. It makes my day brighter. Theresa has to remember that I have lived on my own for a very long time. And now here I am, trying to fit in with a family of sorts. No one can do this overnight, autistic or otherwise. The past 3 years of my life have been filled with nothing but conference appearances, because I have been very depressed. I find I am not so depressed when I am with her, but now I have the problem of figuring how to fit in and join a family unit, as opposed to being on my own.

The nice thing about living alone is that you can do what you want when you want to do it. I have found that this is not so in Bedford. Even something so simple as checking e-mail is a challenge when the kids want your attention. You have to learn to balance your own needs and desires with those of the children you are helping to raise. This is not an easy thing to do, but I am trying to learn, because I have felt a sense of accomplishment when I get it right.

There is also a problem of sensory overload that is evident in autism. With three kids, this happens to me more than she thinks it does. I think it happens to her more than she thinks it does. We are still working on how to deal with this, and it has led to many arguments. But one thing I can say about Theresa is that we always make up.

Theresa
As Thanksgiving is almost here, we have found another issue that is becoming more apparent to us–socialization. Thomas is terrified about going over to my parents for Thanksgiving. This has caused a huge disagreement between us. I have felt that you need to spend holidays with your family and that it is a time for celebration. Thomas informed me a few days before the holidays that he did not like holidays and that he would prefer to stay at home while we went to my family’s house for dinner. I am well aware of the anxiety that Thomas would have to go through to attend holidays with others, but I offered some compromises, such as a room he could escape to that had a VCR and a TV in it as well as a bed so that he could have some down time. This still is causing some stress between us, and it is making me realize that there are many things that will not be “typical” in our lives. Thomas is going to try. Maybe he can bring his car so that if it gets to be too much, he can go home. Again, his willingness to try has been a strength for us.

Thomas
Our lives will never be “typical,” there can be no doubt about that. That can be said to make it more exciting, though. Never a boring moment with Theresa. Not like there was when I was alone. It is no secret to those who know me that I don’t like holidays. I can be bribed to do Christmas, and I usually am bribed by these things called “presents,” but other than that, I’d be happy leaving the holidays alone because all they do for me is remind me how different I am, and I can surely do without that. Thus, spending a holiday with a family that I don’t even know is not something that I find to be comfortable. Spending a holiday with a family I do know isn’t comfortable, either. We all muddle through these things we don’t like for other people, I guess. The sacrifices of love. Looking at it objectively, how bad can it be?

Theresa
There are many issues on both sides of the fence. I know that for Thomas my three children and my trust and anxiety issues about going into another relationship are a concern. Thomas has been great with the kids; there have been many times when he comes to me and asks what to do in a certain situation. There are times I have had to interrupt and mention different ways he could have handled a problem that arose. I have felt that our communication skills have gotten us through a lot of these obstacles. When it does become too much for Thomas, he always has an escape route-somewhere he can go to take a break away from everything. There is something to be said about having children with autism. The environment is very conducive to any child or adult who falls into that spectrum. I am aware of the surroundings not becoming too overstimulating and of how to handle behavior issues as they arise. I feel like I am an expert in sensory integration methods. Thomas brings wonderful insight and a tremendous source of encouragement to me and the kids. There are many positive things in our relationship. One of them is that he continues to help and understand the kids like no one else can.

Thomas
Theresa’s trust and anxiety issues are very hard for me. She likes to compare me to her previous relationship, and she wonders if I will hurt her. I tell her I love her, and she thinks I am making it up. That really hurts me sometimes. I have no intention of hurting her because I love her. She has suggested that we go to therapy sessions together as a couple. While most men would reject this idea, I actually welcome it, as it will give the relationship a better chance to succeed. If I am going to move from where I have lived all my life to someplace I have not lived before, I want–even need–all the help I can get for this relationship to succeed.

Theresa has shown me that I have knowledge of myself I didn’t know was there. She has shown me that I am stronger than I think I am. She has shown me that I do have talents that have been going to waste for the past few years in that apartment. She motivates me to move forward, one small and slow step at a time.

Theresa
There are many ways I have adapted to make Thomas’ “world” a little better. Thomas is disorganized, and he tends to forget where he puts very important things like his medication or his sunglasses. Thomas hates bright sunlight, so I always make sure to have his sunglasses ready for him. If we are heading out, I put them in my purse when we are going somewhere so they are ready before we go out into the sun again. The sun causes him to be physically sick, so I try to manage to make him more comfortable. I know the things that can help him regain his “balls,” so that he is more able to cope with his day. I know big hugs and deep pressure are a way for him to stay focused in his current environment.

Thomas
There are many ways I have adapted to make Theresa’s “world” a little better. The main thing I have done is to take on a father-figure role for the children, especially the older two. I play with them, feed them, put them to bed, and I watch them for her while she is away, without asking for compensation. The oldest, JR, is very easy to like. I can go to a toy store and get a toy that I like, and he will like it, too. Amy will start out being afraid, but she will eventually join us, and that makes it even more fun.

I don’t always like getting out of bed in the morning because I am so used to sleeping through it. Sometimes there is a doctor’s appointment or something going on at the school that I have to be involved with. I have found with time that these things are not only okay, they are very special. This may not be the ideal family, but it is a family that loves me. To me, that is what matters. I am fortunate to have them.

Theresa
As we arc fortunate to have Thomas. Through all of our difficult transitions and the trials we have had to experience, one thing does remain: Our love and commitment to one another has withstood all of our difficulties. It took us a while to make that commitment. There were many times that we wanted to walk away from it and from all the problems that we knew we were going to encounter. We have handled our problems not only by communicating but by praying together. God is something we both have in common, and a tool that we have used because if we did not, we know that our relationship would be impossible. Thomas has been great about going to church, even though the social nature of it at times is too much for him. He has a good relationship with the pastor, and the church has welcomed him with open arms. Once again, Thomas has shown me that he cares enough about us that he is willing to overcome his initial fear and transition problems.

Thomas
One time Theresa and I were at a restaurant, and I wagged a fork at a waitress and Theresa said I was flirting with her. So now I wag my fork at Theresa. It has been very hard getting to where we are, and we have both walked away more than once, only to get back together again. The fact that we do get back together again gives me great hope for our future.

Theresa
I must also realize not to expect so much from him. This is very hard for me to do. I forget that he is not comfortable with some things, that he would prefer certain routines, and that he has his own expectations, not mine. When it does become too much for him, and he reminds me of the struggles he faces, they are, therefore, my struggles as well. I want Thomas to fit into our home and our lives; I want him to feel like my home can be his as well. In order to do this, I must put my own needs and desires aside at times. This is hard for me; I want to also be as independent as possible. As I watch how hard he works, I am compelled to give up things that I would rather do to please him and make him happy. So many times I have seen Thomas do things that I know are so hard for him, so I also have committed to doing things that don’t always make me comfortable. An example of this would be shooting rockets. I do love to do this with him and the kids, but after half an hour, I’ve had it. I have learned, however, that this is a way that Thomas gets his “balls” back, and when he is happy and content, I am as well.

Thomas
The rockets are fun mainly because of the kids. It is really fun to see a rocket go up a thousand feet or so and to have one of the kids catch it on the way down. I also like it because it is fun for me as well. I like to do things like fire rockets and fireworks and ride the go-carts and such. I never really had a chance to do that when I was younger, so I am making up for that now. I am very glad that Theresa understands this.

Just as she must put her own needs aside sometimes, I must do that, too. The kids sometimes–many times–come first. This is by nature of having children, and it is something that I am not so sure is bad, but it will take some getting used to.

Theresa
Our roller coaster goes up and down a lot of days. When Thomas does not seem to understand a social situation, or when it becomes too much for him, he tends to shut down. This is the most difficult time for me because many times I do not know how to fix it. He just runs off somewhere and I get worried; I know he needs that time, but I wonder how long it will be and how I can help. He can usually shrug it off eventually, but I feel useless at those times in our lives. Thomas also can be a little impulsive and do or say things and later realize it was not the most socially appropriate thing to do. He has been very good about apologizing at those times. He is also very good about asking and restating what the most “typical” thing to do in that situation would be. There has been many a time I have wanted to hide when Thomas does this. What I have learned is that Thomas really has to be taught all of those social graces. We tend to take this for granted, but he has to learn what to do, memorize it, and then try to replicate it again in a similar situation. Thomas tends to overexaggerate it at times. The constant practice at times can get to me.

We role-play quite a bit to achieve these social graces. Thomas loves to role-play to make sure he gets it right, so he feels like he fits in. A neat example of this was a few months ago, when I mentioned to Thomas that not everything needed a verbal answer; sometimes a simple gesture would do. Thomas went right to work starting to figure out what gestures were appropriate in certain situations. He would ask others who were close to him, “Was that a good gesture? Was that right?” It was funny, but I realized what I had taken for granted; this man must literally practice daily. We also role-play to make sure the right terms and meanings come out. Thomas will say almost daily, “Let’s try that conversation again to make sure we have it right.” He has been very open about criticism and always willing to try out new things. Once again, our saving grace has been Thomas’s ability to try almost anything. He wants to make sure he fits in, and he feels bad when he feels that he does not.

Thomas
The role-playing goes both ways. I had heard that people with autism did better when they role-played, so I decided to try it. Theresa has also role-played. An example of this would be a recent IEP meeting for JR. We went to a restaurant, and I pretended to be the bad guy. I role-played some of the things I thought he might say so Theresa could practice what to say back to him. She did very well, and I was proud of her. She did even better at the actual IEP meeting.

I don’t think I will ever understand gestures. I have been practicing and trying to get them right, but they don’t really make sense to me. I watch people on TV do it, and they seem so casual about it. Jason Alexander is a master of gestures. My life is complicated enough; I don’t need to turn into a George Costanza to make it even more difficult.

Social graces are lost on me, too. Nowadays, everything must be “politically correct,” and it drives not only me crazy but many others as well! Another example is when Theresa went through the toll booth; she gave the lady her money and said, “Thank you.” Why would you want to thank someone for taking your money? These are the kinds of things about the general population that I don’t understand. Maybe I don’t understand them because they don’t make sense?

Theresa
What Thomas does not realize at times is that I love him just the way he is, even lacking those social graces. Thomas’ usual wit and charm seem to rescue him many times with other people–as well as his constant impressions of others to relieve any dry moment in conversations. Thomas is a good man with a very compassionate heart. He shows his feelings in a way that I have seen no other man do. The movies that make me cry make him cry. The sadness I feel and verbalize he feels and verbalizes. These are great traits that show how open and compassionate he truly is. I would not know what to do without his wonderful hand that holds mine or without his hugs that make me feel like everything will be OK. He has a great saying that he uses and that we worked on during one very trying time in my life. He just says over and over, “Not to worry, everything is under control,” and I slowly settle down and believe it. He just lightly strokes my head as he says it, and I know he has taught me as much as I have taught him. He has taught me to slow down to appreciate those things that I take for granted–that a long hug is worth the world stopping for a short time and that it is still OK to go out and play every day.

Thomas
It is hard making it work. Yet at the same time, I feel a great peace when I am with her. This is something that is new to me. I like it. I’d also like to think I deserve it. She is a strong and wonderful woman. I am glad I met her. I am a better man for the journey.

Theresa
Yes, we will make it work. We know the struggles that we have to face–too many for me to even think about at this time. As we take one day at a time, we know that it will be worth all the work and all the time and energy a relationship takes. Thomas sings me a song that he has written called Always Wanted. This song describes the perfect woman and how he has always wanted her. The last verse is my favorite. After explaining this perfect woman, he says quietly, “You are what I always wanted.” It is very touching to me. What he does not realize is that he is what I have always wanted, too.

Youth dating violence

ABSTRACT
Adolescents’ responses to the Youth Dating Violence Survey have previously been documented by the Centers for Disease Control and Prevention (1996). The present study on dating violence examined the responses of thirty-seven adolescents enrolled in an alternative high school program. Many reported psychological victimization in a dating relationship: their partners did something to make them feel jealous, damaged their possessions, said things to hurt their feelings, insulted them in front of others, tried to control them, threatened them, blamed them for bad things the dating partners did, and brought up something from the past to hurt them. In terms of perpetrating psychological abuse in a dating relationship, over half of the adolescents reported that they hurt their dating partners’ feelings, insulted them in front of others, did something just to make them jealous, tried to control them, and damaged their possessions. Many of the adolescents had also been victims of physical violence in their dating relationships; they reported being scratched, slapped, slammed or held against a wall, kicked, bitten, forced to have sex, choked, and pushed, grabbed, or shoved, as well as having their arms twisted and fingers bent. Some perpetrated physical violence in dating situations, such as scratching their dating partners, hitting them with a fist or something hard, throwing something that hit their dating partners, kicking them, slapping them, physically twisting their arms, slamming or holding them against a wall, bending their fingers, biting them, choking them, and pushing, grabbing, or shoving them. The findings confirm that dating violence among adolescents is a serious health problem that needs to be addressed.

The majority of research on dating violence has focused on college students, a population that is not very representative of high school students and ethnic minority groups. However, researchers increasingly are examining intimate partner violence using high school samples (Barth & Derezotes, 1990; Cascardi, Avery-Leaf, & O’Leary, 1994; Foshee et al., 1996; Jaffe et al., 1992; Sundermann & Jaffe, 1993).

In their groundbreaking study of high school age youth, Henton and Cate (1983) found that 12% had experienced abuse in one of their dating relationships. Roscoe and Callahan (1985), examining middle-class high school students in a white Michigan community, found that 9% of the males and females had experienced physical violence while on a date, with 10% of the females reporting physical violence in a dating relationship. Among adolescents from both predominantly white suburban and rural schools and multiracial inner-city schools, it was found that approximately 20% had experienced violence from a dating partner (Bergman, 1992). Additional studies on adolescents experiencing at least one incident of physical violence in a dating relationship have reported the following rates: 19% (Roscoe & Kelsey, 1986), 27% (O’Keefe, Nona, Brockopp, & Chew, 1986), and 38% (Molidor & Tolman, 1995). Jezl et al. (1996), examining abuse in a coed, ethnically diverse, religiously affiliated high school sample, found that 59% had been the victim of physical violence, 96% had experienced some form of psychological abuse, and 15% had been forced to engage in sexual activity at least once in a past or current dating relationship. Foshee et al. (1996) investigated a racially diverse sample of adolescents (75.9% Caucasian, 20.2% African American, and 3.9% other racial/ethnic groups) and found that approximately 20% had experienced dating violence. O’Keefe’s (1997) research on high school students (53% Latino, 20% white, 13% African American, 6.7% Asian American, and 7% other racial/ethnic groups) revealed that 43% of the females and 39% of the males had been physically aggressive toward a dating partner at least once.

These findings indicate that dating violence is a significant problem among adolescents. Thus, influences, contexts, and other relevant variables need to be further explored and clarified. The present study focused on the responses of high school students to questions on the Youth Dating Violence Survey (Centers for Disease Control and Prevention, 1996), pertaining to psychological and physical abuse in dating relationships.

METHOD
Data were obtained from 37 adolescents (17 males and 20 females) who were enrolled in an alternative high school program in the Pacific Northwest. They ranged in age from 14 to 18 years (the average age was 16.58 for females and 16.95 for males). Ten were of mixed ethnic heritage, 10 were white, 8 were African American, 7 were Hispanic/Latino, 1 was Asian, and 1 did not report ethnicity. Ninety-five percent of the adolescents had a mother at home, 31% had a father at home, and 29% were living with either both mother and father, mother and stepfather, or two foster parents.

The adolescents were administered the Youth Dating Violence Survey (Centers for Disease Control and Prevention, 1996), which gathers information on demographics, dating behavior, and mediating variables. Examples of items pertaining to dating violence include the following: “It is OK for a boy to hit his girlfriend if she did something to make him mad” and “Boys sometimes deserve to be hit by the girls they date” (response categories: strongly agree, agree somewhat, disagree somewhat, or strongly disagree). The following are examples of items measuring conflict management skills: “I threw something at the person I was mad at” and “I told the person why I was angry” (response categories: very often, sometimes, not very often, or never). Those adolescents who had ever been on a date were asked about psychological and physical victimization or perpetration: “How often has anyone that you have ever been on a date with done the following things to you?” and “How many times have you done the following things to anyone that you have been on a date with?” (see the list of behaviors appearing in the tables in the Results section), with different response categories for psychological abuse (very often, sometimes, seldom, or never) and physical abuse (10 or more times, 4 to 9 times, 1 to 3 times, or never). Additional data were collected on awareness of services for victims and perpetrators, help-seeking behaviors of victims and perpetrators, and belief in the need for the victim or the perpetrator to get help.

Parental consent was obtained from 42 of the 47 eligible adolescents, or 89%. Surveys were completed by 37 of these 42 adolescents, or 88% of those whose parents gave consent (5 adolescents did not complete the survey because they either were absent the day of data collection, did not answer all of the questions, or had reading difficulties that precluded completion of the survey during the allotted time). The survey was administered in accordance with local school district and university institutional review board policies (with special provisions related to school-based populations), and took approximately forty-five minutes to complete. In some instances, items were read aloud due to concerns about participants’ reading and comprehension levels. They were assured that their responses would remain confidential.

Most of the adolescents had been victims of psychological abuse in one form or another in their dating relationships (see Table 1). Over three quarters reported that their dating partners did something to make them jealous. Approximately half indicated that their dating partners would not let them do things with other people, threatened to start dating someone else, told them they could not talk to someone of the opposite sex, brought up something from the past to hurt them, and blamed them for bad things the partners did.

Table 2 presents the findings regarding adolescents’ perpetration of psychological abuse. Over three quarters reported that they said things to hurt their dating partners’ feelings on purpose. Approximately half said they damaged something that belonged to their dating partners, insulted them in front of others, threw something at them but missed, would not let them do things with other people, told them they could not talk with someone of the opposite sex, did something just to make them jealous, and brought up something from the past to hurt them.

Many of the adolescents had been victims of physical violence in their dating relationships (see Table 3). From approximately a third to a half reported being scratched, slapped, slammed or held against a wall, kicked, bitten, forced to have sex, choked, and pushed, grabbed, or shoved, as well as having their arms physically twisted and fingers bent.

Table 4 indicates that some of the adolescents also perpetrated physical violence in dating situations. Approximately half reported scratching their dating partners, hitting them with a fist, and pushing, grabbing, or shoving them (around a quarter reported hitting their dating partners with a fist and pushing, grabbing, or shoving them 4 to 9 times). A third or more threw something that hit their dating partners, kicked them, and slapped them. About a fifth reported physically twisting their dating partners’ arms, slamming or holding them against a wall, bending their fingers, biting them, choking them, dumping them out of a car, burning them, beating them up, and hitting them with something hard besides a fist.

DISCUSSION
The findings regarding physical and psychological aggression in dating relationships are consistent with those of previous studies. Cascardi, Avery-Leaf, and O’Leary (1994) found that 32% of the males and 52% of the females in their study of high school students had used aggression against a dating partner, while O’Keefe (1997) found that 39% of males and 43% of females had been physically aggressive with a dating partner at least once. Additionally, Foshee (1996) reported that females were more likely to be perpetrators of dating violence. Jezl, Molidor, and Wright (1996) found that 59% of their adolescent sample had experienced physical violence at least once in a dating relationship, 96% had experienced some form of psychological maltreatment, and 15% had been forced to engage in sexual activity. Significantly more males than females reported being victims of physical abuse.

The present study did not include an analysis of different types of dating profiles: (a) victim only, (b) perpetrator only, or (c) mutually violent relationship. In this regard, it is worth noting that adolescents in mutually violent relationships have been found to receive and perpetrate significantly more abuse as compared with adolescents in one-sided violent relationships (Gray & Foshee, 1997). Further research on these dating violence profiles is needed, and gender and ethnic group differences should be analyzed.

O’Keefe (1997) reported that high school males were more likely to harm a dating partner if they had witnessed interparental violence, believed that male-female dating violence was justifiable, used alcohol or other drugs, were the recipients of dating violence, and had experienced more conflict in their dating relationships. Females were more likely to be violent toward a dating partner if they believed that female-male violence was justifiable but male-female violence was not, had experienced more conflict in their dating relationships, were the recipients of dating violence, used alcohol or other drugs, and felt the relationship was more serious. O’Keefe noted contextual correlates, such as experiencing physical aggression, interparental aggression, acceptance of violence, history of aggression, and community and school violence. Additional research by O’Keefe (1998) indicated that, among adolescent males who witnessed high levels of interparental violence, those who abused their dating partners were differentiated from those who had violence-free relationships by the following variables: low socioeconomic status, exposure to community and school violence, acceptance of violence in dating relationships, and low self-esteem. Low socioeconomic status and acceptance of violence in dating relationships differentiated males who experienced dating violence and those who did not. For females, exposure to community and school violence, poor school performance, and experiencing child abuse differentiated those who abused their dating partners and those who did not, while poor school performance and experiencing child abuse differentiated females who experienced dating violence and those who did not. The present research would have benefited from the inclusion of such factors as socioeconomic status, acceptance of violence in dating relationships, self-esteem, school performance, child abuse, and exposure to family, community, and school violence. Nearly all of the adolescents in this study lived in an area where exposure to violence was not unusual.

Overall, the literature indicates that intimate partner violence during adolescence is a serious problem. The findings here reinforce the notion that at least 25% of adolescents experience psychological and physical abuse in their dating relationships. Romance and violence are increasingly gaining acceptance as a version of love and war among teenage dating couples. As greater understanding of the factors associated with dating violence is achieved, more effective prevention and intervention programs can be designed and implemented.

Gray and Foshee (1997) found that about 86% of the adolescents who reported violence in dating relationships stated that it was mutual. Moreover, social learning theory would indicate that individuals in such relationships may escalate the violence. Thus, future research should investigate the characteristics of mutually violent relationships. The following questions should be addressed in this regard: Do violent adolescents seek out dating partners who are also violent? Does the violence escalate as a result of duration, frequency, and exclusivity of dating? It is also important to evaluate age, gender, ethnic background, family history, educational level, and other factors that might mitigate, or contribute to, adolescent dating violence.

The results indicated that males and females perpetrated psychological abuse and were victimized at about the same rate, but that males engaged in higher levels of physical violence in dating relationships. This finding is consistent with past studies of dating violence (Foshee, 1996).

Prevention programs focused on adolescent dating violence (see Foshee & Langwick, 1994; Foshee et al., 1998) need to include lessons and role-playing related to one-sided and mutually violent relationships (e.g., discussions should cover the characteristics and consequences of such relationships). Health providers and educators need to ensure that adolescents are taught social skills to replace the use of psychologically and physically abusive, controlling behaviors in dating relationships, and are given opportunities to reappraise beliefs about gender roles.

The low number of adolescents in the sample limits the generalizability of the findings. Nevertheless, the results contribute to our understanding of the types and prevalence of dating violence among adolescents. The etiology and course of violent dating relationships should be a priority of future research. Findings will serve to provide a framework for the prevention of intimate partner violence among adolescents, a serious health problem that urgently needs to be addressed.