JOURNAL OF SEX RESEARCH, 48(5), 450460, 2011 Copyright # The Society for the Scientic Study of Sexuality ISSN:

0022-4499 print=1559-8519 online DOI: 10.1080/00224499.2010.509892

Early Age of First Sexual Intercourse and Depressive Symptomatology among Adolescents
Luanne K. Jamieson and Terrance J. Wade
Department of Community Health Sciences, Brock University Past literature has provided conicting evidence for the association between adolescent sexual intercourse and depressive symptomatology. Whereas some studies conclude that sexually active youth may be at risk for depression, others provide contrary results. Thus, it is unclear as to whether depression results directly from coitus or if this relationship is explained by other factorsthat is, there may be biological, psychological, or sociological predictors of both depressive symptomatology and early sexual intercourse. Using the National Longitudinal Study of Adolescent Health dataset, depressive symptomatology in adolescents over a seven-year time period was analyzed. The nal sample (n 6,510) was comprised of 49.35% male (n 3,213) and 50.65% female (n 3,297) participants. Although an earlier age of rst coitus was predictive of future depressive symptoms, both variables appear to be concomitant outcomes of the biopsychosocial process. Thus, although one may be able to use early coitus as a marker for subsequent depressive symptomatology, it does not appear to occur because of early sexual intercourse. Furthermore, the reverse relationship was not found to be signicant in this studythat is, higher levels of previous depressive symptomatology did not predict an earlier age of rst sexual intercourse in adolescents.

Depression and sexual intercourse are relevant to todays youth in that adolescence is often the time when depression rst appears (Kaltiala-Heino, Kosunen, & Rimpela, 2003; Robins, Locke, & Regier, 1991; Scoureld et al., 2003; Wade, Cairney, & Pevalin, 2002) and sexual intercourse begins (Centers for Disease Control, 2007; Smith, Guthrie, & Oakley, 2005). Past research suggests an unclear relationship between earlier sexual intercourse and depression. Whereas some studies conclude that sexually active youth may be at greater risk for depression (Hallfors et al., 2004; Rector, Johnson, & Noyes, 2003; Tubman, Windle, & Windle,

This research uses data from the National Longitudinal Study of Adolescent Health (Add Health), a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by Grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. No direct support was received from Grant P01-HD31921 for this analysis. We thank Drs. Anthony F. Bogaert, Stan W. Sadava, and B. J. Rye for their suggestions on a previous draft of this article. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data les from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin St., Chapel Hill, NC 27516-2524 (addhealth@unc.edu). Correspondence should be addressed to Terrance J. Wade, Department of Community Health Sciences, Brock University, 500 Glenridge Ave., St. Catharines, Ontario L2S 3A1, Canada. E-mail: twade@ brocku.ca

1996), others provide contrary results (Lehrer, Shrier, Gortmaker, & Buka, 2006; Sabia, 2006) or ndings that are only representative of high-risk sexual behaviors, such as intercourse without a condom (Hallfors, Waller, Bauer, Ford, & Halpern, 2005). These contradictory ndings suggest that the relationship between sexual intercourse and depression may be explained by other factorsthat is, there may be biological, psychological, or sociological variables that may predict both. Meiers (2007) research suggests that age of rst sexual intercourse is not signicantly related to future mental distress, except as a marker of previous factors. Thus, by changing sexual behavior, one would not prevent potential mental distress. Results of the six previous studies have various limitations, including the use of cross-sectional data, longitudinal results that are separated by only one or two years, use of one item from a multi-item scale to analyze depression, and insufcient data on what represents early sexual intercourse. We address these limitations in this study by longitudinally examining this relationship over a seven-year time interval to assess its directionality using the National Longitudinal Study of Adolescent Health (Add Health; Harris, 2009). The Add Health is a nationally representative, school-based study of adolescents who were in Grades 7 through 12 in the United States (including Alaska and Hawaii) in the 1994 through 1995 academic school year. Data were

SEXUAL INTERCOURSE AND DEPRESSIVE SYMPTOMATOLOGY

collected from the students and parents in the 1994 through 1995 academic year (Wave 1) and from the adolescents in the 1995 through 1996 academic year (Wave 2). A third wave of data was collected from the respondents, as young adults, in 2001 through 2002 (Wave 3).

Research Questions The contradictory ndings of previous studies suggest that the relationship between early coitus and depressive symptomatology may be explained by biological (age, sex, pubertal maturity, and race), psychological (selfesteem and social support), and sociological (socioeconomic status [SES] and religious involvement) predictors. By grounding the analyses within a biopsychosocial framework (Engel, 1977), we begin to assess the factors that may predict both sexual intercourse and depressive symptomatology as concomitant outcomes of a similar process. Figure 1 illustrates the possible links between adolescent depressive symptoms, sexual intercourse, and the potential biopsychosocial predictors. From this gure, and using the Add Health dataset, we derived four specic research questions:
RQ1: Do biological, psychological, and sociological factors at Wave 1 predict an earlier age of rst sexual intercourse (Waves 1, 2, or 3)? (Figure 1, Q1) a. Does a higher level of depressive symptomatology at Wave 1 predict an earlier age of rst sexual intercourse (Waves 1, 2, or 3)? (Figure 1, Q1SA1) RQ2: Do biological, psychological, and sociological factors in Wave 1 predict higher levels of depressive symptomatology (and an increase in depressive symptomatology) in Wave 3? (Figure 1, Q2)

RQ3: Does an earlier age of rst sexual intercourse (Waves 1, 2, or 3) predict higher levels of subsequent depressive symptomatology (and an increase in depressive symptomatology) at Wave 3? (Figure 1, Q3) RQ4: After controlling for biological, psychological, and sociological factors at Wave 1, does earlier age of rst sexual intercourse (Waves 1, 2, or 3) predict higher levels of depressive symptomatology (and an increase in depressive symptomatology) in Wave 3? (Figure 1, Q4)

These questions determine if both age of rst sexual intercourse and depressive symptomatology are predicted by similar biopsychosocial variables.

Method Participants An overview of the Add Health sampling information and research design was provided by Chantala and Tabor (1999) and Harris et al. (2009). A stratied, random sample of schools were selected, which included 80 high schools and 52 of their afliated middle schools. Students were randomly sampled from schools and regions. African Americans, Chinese, Cuban, Puerto Rican, and disabled adolescents were over-sampled to provide better estimates for these specic subpopulations. Sensitive questions (on topics such as sexual activity and depressive symptomatology) were entered by the students themselves using an audio computer-assisted self-interview (A-CASI) device. The A-CASI interview technique has been shown to increase the validity of sensitive question reporting (Supple, Aquilino, & Wright, 1999) while maximizing condentiality, improving authenticity (Turner et al., 1998), and reducing missing data (Hallfors, Khatapoush, Kadushin, Watson, & Saxe, 2000). Measures Depressive symptomatology. Depressive symptomatology was consistently assessed in all three waves of the Add Health utilizing eight items of the 20-item Center for Epidemiological StudiesDepression (CESD) Scale (see the Appendix). Developed by Radloff in 1977 to be standardized for epidemiological studies and high school populations (Radloff, 1991), the CESD assesses current levels of depressive symptomatology by asking respondents to rate various depressive symptoms over the last seven days. Response choices are 0 (never or rarely) to 3 (most of the time or all of the time). To create the measure of depressive symptomatology utilized in this study, all eight CESD items were summed across each wave. Scores on the shortened version of the CESD ranged from 0 to 24, with higher values on this scale representing higher levels of depressive symptomatology. A factor 451

Figure 1. Proposed model of the association between age of rst sexual intercourse and depressive symptomatology, which includes biopsychosocial factors.

JAMIESON AND WADE

analysis using the maximum likelihood method with a principal components and a varimax rotation solution was conducted and veried that these eight items demonstrated similar factor structures to previous analyses (Radloff, 1977; Roberts, Andrews, Lewinsohn, & Hops, 1990). Cronbachs alpha for these eight items was 0.79 (Wave 1), 0.80 (Wave 2), and 0.81 (Wave 3). For respondents who did not answer one or more of the CESD items for any wave, a respondent-based mean substitution technique was employed using the individuals responses to items (Anderson, Basilevsky, & Hum, 1983). Respondents who had missing data on less than 30% of items on this scale in a specic wave were assigned the average score of the remaining items of the scale in that wave. Those who were missing more than 30% of the items were deleted from analysis. The use of a respondent-based mean substitution technique assumes that the mean across the items answered by the respondent is representative of the average of all items in the scale. This technique is preferable to a sample mean substitution technique because it does not assume that each case can be adequately represented by the average of all respondents, and it does not articially reduce the variance of the sample (Anderson et al., 1983). Age of rst sexual intercourse. The Add Health dened sexual intercourse and vaginal intercourse as malefemale penilevaginal intercourse. Sexual intercourse was assessed by self-report in Waves 1 and 2 by the question, Have you ever had sexual intercourse?, and in Wave 3 by the question, Have you ever had vaginal intercourse? Those respondents who answered yes to these questions were further prompted as to when rst intercourse occurred. Respondents year of birth was then subtracted from the reported year of rst sexual intercourse to provide an age of rst coitus. Earlier adolescent sexual intercourse was dened in this study as malefemale penilevaginal intercourse between ages 11 and 16. All three waves were used to identify age of rst sexual experience because during Waves 1 and 2, many respondents, especially the younger ones, had reported never having sexual intercourse. Wave 3 provided an opportunity to minimize missing data due to this right-hand censoring as a result of young age because all respondents were at least 18 at this time, and most had reported a rst sexual experience involving coitus by this time. Right-hand censoring refers to the occurence of an event after the last observation time in the study has been recorded. Age at rst sexual intercourse was coded so that Wave 1 data were utilized in analyses. If Wave 1 data were missing, then Wave 2 data were used. If both Wave 1 and Wave 2 data were missing, then Wave 3 data were used. Thus, age of rst vaginal intercourse was computed by subtracting the reported age of rst coitus (Waves 1, 2, and 3) from the corresponding year in which the respondent was born (Waves 452

Figure 2. Frequency of age of rst sexual intercourse among adolescents (n 6,510).

1, 2, and 3), creating age of rst sexual intercourse as a self-reported continuous variable coded in years. Selfreport of sexual behavior has been shown to be consistent with actual behavior (see Alexander, Somereld, Ensminger, Johnson, & Kim, 1993). In some instances, when responses to these questions were contradictory by one participant over multiple waves, a decision was made to utilize their responses from Wave 3. The logic behind this decision was that there might be fewer exaggerated responses among the respondents when they were older, and, as such, Wave 3 may provide a more accurate date of rst coitus. When there was a discrepancy between waves on the reported age of rst coitus, the respondents Wave 3 answer was utilized. Based on this, age of rst sexual intercourse for the nal sample ranged from 11 to 25 years of age (see Figure 2). Two exclusions for this variable included age of sexual intercourse prior to 11 years of age and same-sex intercourse. Sexual intercourse before 11 years of age was considered to be nonconsensual, and would likely be representative of forced sexual intercourse or sexual abuse. Childhood sexual abuse is associated with later depression (Molnar, Buka, & Kessler, 2001), and would likely bias the nal results. Thus, these participants were removed from analyses. Second, due to the independent associations for same-sex attraction and depression (Lam et al., 2004), this study did not include sexual intercourse with same-sex partners. Respondents (n 71) who identied as homosexuals (100% homosexual or mostly homosexual, but somewhat attracted to people of the opposite sex) or asexuals (not sexually attracted to either males or females; for a comprehensive overview, see Bogaert, 2004, 2006, 2008) were removed from analyses. Biological Variables Age. Age was coded by year and was calculated during Wave 1 by subtracting the date of birth from the date that the rst interview was completed. The age range for adolescents in this study in Wave 1 was 11 to 21 years.

SEXUAL INTERCOURSE AND DEPRESSIVE SYMPTOMATOLOGY

Because age changes at a constant rate among individuals across each wave, only age at Wave 1 was used. Sex. Sex was a self-reported variable that was conrmed by the interviewer (by asking, if necessary). Males were coded as the referent group for all analyses (male 0, female 1). The nal sample included 3,297 females and 3,213 males. Pubertal maturity. In Wave 1, relative physical development was assessed in relation to other adolescents. Respondents were asked, How advanced is your physical development compared to other girls=boys your age? Responses ranged from 1 (younger than most) to 5 (older than most). Because pubertal maturity was based on subjective self-report, an objective measure of maturity (age of menarche) was added to the analysis for females. Age of menarche. Menarche was assessed at Wave 1 by the question, How old were you when you had your very rst menstrual period? The sample mean age of rst menstrual cycle (X 12.09) was then subtracted from the respondents age of rst menstruation. Because this is applicable only to females, a conditionally relevant variable was created to allow the examination of age of menarche for females while maintaining males in the analysis. Race. Race was based on respondents self-report at Wave 1. Respondents were asked to identify their race based on the following four options: White, Black or African American, American Indian or Native American, and Asian or Pacic Islander. Each answer was dichotomized as 1 yes and 0 no. Race was then recoded into four dichotomous variables, with White as the referent group. Psychological Variables Self-esteem. Self-esteem was determined in Wave 1 using a six-item subset of Rosenbergs Self-Esteem Inventory, which is a measure of global self-esteem (Rosenberg, 1965). Responses were based on a ve-point Likert scale that ranged from 1 (strongly agree) to 5 (strongly disagree). Answers were reversecoded so that they ranged from 6 to 30, with higher values representing higher levels of self-esteem. A principal components factor analysis, using the maximum likelihood method with varimax rotation, indicated a single-factor solution. The scale was built by summing all six items and utilizing the same missing value algorithm previously described for the CESD. Cronbachs alpha for this scale was 0.85. Social support. This variable was assessed based on responses to ve Wave 1 items (e.g., How much do you

feel that your friends care about you?, and How much do you feel that people in your family understand you?). Responses ranged from 1 (not at all) to 5 (very much). Social support items were summed to create a measure of social support ranging from 5 to 25, with increasing values indicating higher social support. The same missing value algorithm was employed in the construction of this measure as in the previous scales. Sociological Variables Socioeconomic status (SES). SES was assessed using parent-reported household income and household education collected in Wave 1. Income was measured using the following: About how much total income, before taxes, did your family receive in 1994? Include your own income, the income of everyone else in your household, and income from welfare benets, dividends, and all other sources. For analysis, household income was divided by 1,000 so it ranged from 0 to 999, and was examined as a continuous variable in thousanddollar increments. Household education was assessed by the questions, How far in school did you go?, and How far did your current spouse=partner go in school? Answers ranged from 1 (never went to school) to 10 (professional training beyond a four-year college or university degree). To assess household education and compensate for singleparent households, the highest education obtained by either the respondents mother or father was utilized. A series of dummy variables were created to compare each category of education to the reference category, Graduated from a college or university. Religious involvement. Religious involvement was based on respondents answers to the Wave 1 question, In the past 12 months, how often did you attend religious services? Potential responses included once a week or more, less than once a month, randomly, and never. A series of three dichotomous variables were created to compare each type of religious involvement to never attended religious services as the referent group.

Results To determine the relationship of the biopsychosocial variables of interest on depressive symptomatology and early sexual intercourse, secondary data analysis was conducted on the three-wave Add Health restricted dataset. All statistical analyses were performed using Statistical Analysis Software (SAS) Version 9.1 (SAS Institute Inc., Cary, NC). Because the Add Health was intended to be nationally representative of the population of the United States, longitudinal sample weights were used in all analyses to provide estimates generalizable to the 453

JAMIESON AND WADE

U.S. adolescent population. In addition, all tests of signicance employed a specic regression technique in SAS (PROC SURVEYREG) to correct for the complex sample design of the Add Health to ensure unbiased standard errors. This procedure takes into account the clustering in the sample and the longitudinal weighting to maintain unbiased estimates and standard errors in a way recommended by the originators of the data (see Chantala & Tabor, 1999). With the exception of the individual mean substitution algorithm for depressive symptomatology, self-esteem and social support, cases with missing and out-of-range values were recoded as missing and deleted from further analysis. The nal weighted sample of n 6,510 respondents from Waves 3s, 1, 2, and 3 was utilized for analysis. RQ1 Using data from Wave 1, age of rst sexual intercourse was separately regressed on the biological, psychological, and sociological groups of variables and then simultaneously on all of the biopsychosocial variables combined. This process allowed for the examination of the independent contribution of each domain on age of rst coitus, as well as the adjusted effect of each domain after controlling for the other two domains. Biology (as assessed by age, sex, race, pubertal maturity, and age of menarche) accounted for 6.19% of the variance in age of rst sexual intercourse, F(7, 128) 27.05, p < .001 (see Table 1, Model 1). In Table 1, Model 2, the psychological variables (comprised of self-esteem and social support) accounted for 1.68% of the variance

in age of rst coitus, F(2, 128) 26.53, p < .001. Sociological variables (as assessed by household education, household income, and religious involvement) accounted for 4.37% of the total variance in age of rst coitus, F(5, 128) 25.60, p < .001 (see Table 1, Model 3). When all biopsychosocial variables were included in the model (see Table 1, Model 4), they accounted for 11.94% of the variance in age of rst coitus, F(14, 128) 24.85, p < .001. In sum, the biological variables analyzed were the strongest predictors of age of rst coitus. However, the inclusion of psychological and sociological variables accounted for an additional 5.75% of the variance in age of rst sexual intercourse, almost doubling the explained variance in age of rst coitus compared to biology alone. Sub-analyses. Two additional sub-analyses were conducted using age of rst sexual intercourse and depressive symptomatology. The rst sub-analysis examined only those respondents who reported not yet having their rst sexual intercourse prior to Wave 1 (but would have, or did have, sexual intercourse after Wave 1). This analysis addressed the potential that, among some respondents, psychological and sociological variables were assessed after ones rst sexual occurrence and may be inuenced by this experience. In the rst sub-analysis (see Table 2, Model 1), the biopsychosocial variables predicted 22.63% of the variance in age of rst sexual intercourse among adolescents with no previous sexual debut (n 3,965), F(14, 128) 37.94, p < .001. The second sub-analysis, using the same sample of respondents who reported not yet having had coitus,

Table 1.

Wave 1 Biopsychosocial Predictors of Age of First Sexual Intercourse

Model 1: Biologicala Model 2: Psychologicalb b SE t Model 3: Sociologicalc b SE t Model 4: Biopsychosociald b 0.203 0.052 0.809 0.006 0.347 0.222 0.216 0.005 0.088 0.088 0.002 0.736 0.199 0.076 SE 0.025 0.074 0.114 0.169 0.195 0.035 0.034 0.012 0.016 0.017 0.001 0.096 0.085 0.110 t 8.00 0.70 7.10 0.03 1.78 6.40 6.29 0.45 5.60 5.19 4.19 7.63 2.33 0.69

Variable Age Female Black Native Asian Maturity Age of menarche Self-esteem Social support Household education Household income Church (week) Church (month) Church (random)

b 0.166 0.010 0.789 0.175 0.465 0.226 0.241

SE 0.026 0.077 0.117 0.184 0.217 0.035 0.034

t 6.33 0.14 6.76 0.95 2.14 6.40 7.06

0.001 0.104

0.011 0.017

0.13 6.24 0.096 0.004 0.709 0.193 0.112 0.020 0.001 0.101 0.089 0.109 4.86 5.03 7.04 2.18 1.04

Note. n 6,510 (df 128). Reference groups are male, White, college or university graduate, and never going to religious services. Age of rst sexual intercourse was assessed across all three waves of data collection. a 2 R 0.0619. b 2 R 0.0168. c 2 R 0.0437. d 2 R 0.119. p < .05. p < .001 (two-tailed).

454

SEXUAL INTERCOURSE AND DEPRESSIVE SYMPTOMATOLOGY

Table 2.

Predictors of Age of First Sexual Intercourse in Adolescents with No Previous Sexual Debut
Model 1a Model 2b t 18.93 0.33 3.31 0.64 1.65 3.01 2.65 0.63 3.53 4.71 1.41 4.33 0.95 0.35 b 0.013 0.577 0.025 0.445 0.108 0.301 0.117 0.106 0.008 0.060 0.086 0.001 0.454 0.100 0.043 0.030 0.077 0.134 0.168 0.188 0.039 0.040 0.013 0.017 0.018 0.001 0.105 0.106 0.123 SE 0.012 t 1.11 b 0.013 0.579 0.015 0.436 0.114 0.319 0.117 0.150 0.002 0.058 0.084 0.001 0.453 0.100 0.039 Model 3c SE 0.012 0.031 0.079 0.135 0.168 0.190 0.039 0.040 0.014 0.017 0.018 0.001 0.105 0.106 0.123 t 1.08 18.82 0.19 3.22 0.68 1.68 3.02 2.62 0.17 3.35 4.60 1.37 4.30 0.95 0.32

Variable Depression: Wave 1 Age Female Black Native Asian Maturity Age of menarche Self-esteem Social support Household education Household income Church (week) Church (month) Church (random)

SE

Note. n 3,965 (df 128). Reference groups are male, White, college or university graduate, and never going to religious services. Age of rst sexual intercourse was assessed across all three waves of data collection. a 2 R 0.226. b 2 R 0.001. c 2 R 0.227. p < .01. p < .001 (two-tailed).

was examined with Wave 1 depressive symptomatology as a predictor in addition to the biopsychosocial variables. In this sub-analysis, age of rst coitus was regressed on depressive symptoms rst, and then biological, psychological, and sociological variables were inserted (see Table 2, Models 2 & 3). This analysis determined whether higher levels of prior depressive symptomatology predicted an earlier onset of coitus for those without sexual experience who would have, or did have, sexual experience after Wave 1, and whether this relationship persisted once adjusted for the biopsychosocial factors. Results indicated that Wave 1 depressive symptomatology did not predict subsequent age of rst sexual intercourse among those who had not reported previous intercourse. When age of rst sexual intercourse was regressed on depressive symptomatology and all of the biopsychosocial variables, Wave 1 depressive symptomatology remained nonsignicant (t 1.08, p .281; see Table 2, Model 3). Thus, when separately regressed or when regressed with the biopsychosocial variables, Wave 1 depressive symptomatology did not predict an earlier age of rst sexual intercourse among adolescents with no previous sexual debut. This same null nding was found in sexually active adolescents (see Table 3). RQ2 To assess how biopsychosocial domains predict later depressive symptomatology, Wave 3 depressive symptomatology was separately and simultaneously regressed on the biological, psychological, and sociological variables (see Table 4). Biological variables accounted for 1.89% of

the variance in Wave 3 depressive symptomatology, F(7, 128) 11.31, p < .001. Psychological variables accounted for 4.60% of the variance in depressive symptomatology at Wave 3, F(2, 128) 88.05, p < .001. Both self-esteem (t 10.46, p < .001) and social support (t 2.43, p < .05) were signicant, negative predictors of later depressive symptomatology (see Table 4, Model 2). Finally, sociological variables accounted for 0.80% of the variance in Wave 3 depressive symptomatology, F(5, 128) 7.08, p < .001 (see Table 4, Model 3). When the biopsychosocial variables were simultaneously regressed on Wave 3 depressive symptomatology (see Table 4, Model 4), they accounted for 6.74% of the variance in future depressive symptomatology, F(14, 128) 22.22, p < .001. Thus, only self-esteem and social support appear to be signicant predictors of depressive symptoms. RQ3 To assess whether higher depressive symptomatology was predicted by earlier sexual onset, Wave 3 depressive symptomatology was separately regressed on age of rst sexual intercourse, accounting for 0.564% (t 4.20, p < .001) of the variance in the nal model, F(1, 128) 17.67, p < .001 (see Table 5, Model 1). Thus, every year decrease in rst sexual onset resulted in a 0.126 unit increase in Wave 3 depressive symptomatology. In addition, a sub-analysis examined whether age of rst coitus predicted change in depressive symptomatology at Wave 3, adjusting for baseline depressive symptomatology at Wave 1. This technique partials out the baseline level of depressive symptomatology to explain any changes over time (see Kessler & Greenberg, 1981). 455

JAMIESON AND WADE

Table 3.

Predictors of Age of First Sexual Intercourse in Sexually Active Adolescents

Model 1a Model 2b t 6.19 2.16 0.62 1.17 0.04 3.51 4.42 0.35 1.47 0.15 1.53 2.18 1.36 0.05 b 0.002 0.272 0.191 0.080 0.280 0.010 0.138 0.189 0.005 0.028 0.003 0.002 0.274 0.189 0.007 0.044 0.088 0.130 0.239 0.247 0.048 0.043 0.015 0.019 0.020 0.001 0.126 0.139 0.129 SE 0.010 t 0.16 b 0.007 0.272 0.200 0.089 0.274 0.017 0.167 0.189 0.002 0.029 0.004 0.002 0.273 0.190 0.006 Model 3c SE 0.012 0.044 0.092 0.134 0.240 0.246 0.048 0.043 0.016 0.019 0.020 0.001 0.126 0.139 0.129 t 0.60 6.17 2.15 0.67 1.15 0.07 3.49 4.41 0.11 1.52 0.20 1.55 2.18 1.37 0.04

Variable Depression: Wave 1 Age Female Black Native Asian Maturity Age of menarche Self-esteem Social support Household education Household income Church (week) Church (month) Church (random)

SE

Note. n 2,545 (df 123). Reference groups are male, White, college or university graduate, and never going to religious services. Age of rst sexual intercourse was assessed across all three waves of data collection. a 2 R 0.095. b 2 R 0.000. c 2 R 0.095. p < .05. p < .001 (two-tailed).

When simultaneously regressed to assess change in depressive symptomatology, age of rst sexual intercourse (t 2.98, p < .01) and baseline depressive symptomatology at Wave 1 (t 19.10, p < .001) were both signicant, accounting for 9.71% of the variance in Wave 3 depressive symptomatology, F(2, 128) 184.74, p < .001 (see Table 5, Model 2). Thus, an earlier age of rst coitus signicantly predicted an increase in

subsequent depressive symptomatology after adjusting for baseline depressive symptomatology. RQ4 Depressive symptomatology at Wave 3 was regressed on age of rst sexual intercourse, adjusting for the biological, psychological, and sociological determinants.

Table 4.

Wave 1 Biopsychosocial Predictors of Wave 3 Depressive Symptomatology

Model 1: Biologicala Model 2: Psychologicalb b SE t Model 3: Sociologicalc b SE t Model 4: Biopsychosociald b 0.117 0.454 0.886 0.315 0.292 0.139 0.125 0.184 0.077 0.075 0.001 0.253 0.123 0.189 SE 0.036 0.122 0.174 0.357 0.252 0.051 0.079 0.020 0.029 0.028 0.001 0.166 0.159 0.177 t 3.22 3.71 5.09 0.88 1.16 2.71 1.58 9.36 2.63 2.68 1.32 1.53 0.78 1.07

Variable Age Female Black Native Asian Maturity Age of menarche Self-esteem Social support Household education Household income Church (week) Church (month) Church (random)

b 0.054 0.691 0.751 0.516 0.319 0.147 0.177

SE 0.036 0.112 0.182 0.344 0.245 0.049 0.083

t 1.48 6.17 4.13 1.50 1.30 3.00 2.14

0.193 0.069

0.018 0.029

10.46 2.43 0.106 0.002 0.271 0.144 0.217 0.030 0.001 0.167 0.160 0.183 3.55 2.24 1.63 0.90 1.19

Note. n 6,510 (df 128). Reference groups are male, White, college or university graduate, and never going to religious services. a 2 R 0.019. b 2 R 0.046. c 2 R 0.008. d 2 R 0.067. p < .05. p < .01. p < .001 (two-tailed).

456

SEXUAL INTERCOURSE AND DEPRESSIVE SYMPTOMATOLOGY

Table 5.

Predictors of Wave 3 Depressive Symptomatology

Model 1a Model 2b t 4.20 b 0.082 0.288 SE 0.028 0.015 t 2.98 19.10 0.106 0.451 0.840 0.315 0.311 0.152 0.113 0.184 0.072 0.070 0.001 0.212 0.112 0.184 0.037 0.121 0.172 0.359 0.253 0.051 0.080 0.020 0.029 0.028 0.001 0.171 0.160 0.177 2.83 3.72 4.88 0.88 1.23 2.97 1.41 9.34 2.48 2.50 1.17 1.24 0.70 1.04 b 0.056 Model 3c SE 0.032 t 1.76 b 0.043 0.242 0.153 0.258 0.586 0.315 0.122 0.143 0.101 0.073 0.033 0.031 0.001 0.192 0.112 0.164 Model 4d SE 0.030 0.017 0.036 0.125 0.175 0.328 0.247 0.050 0.078 0.021 0.028 0.027 0.001 0.160 0.157 0.171 t 1.44 14.58 4.29 2.07 3.35 0.96 0.50 2.88 1.30 3.53 1.18 1.14 0.74 1.19 0.71 0.96

Variable Age of rst coitus: Waves 1, 2, & 3 Depression: Wave 1 Age Female Black Native Asian Maturity Age of menarche Self-esteem Social support Household education Household income Church (week) Church (month) Church (random)

b 0.126

SE 0.030

Note. n 6,510 (df 128). Reference groups are male, White, college or university graduate, and never going to religious services. Data are from Wave 1, unless otherwise indicated. a 2 R 0.006. b 2 R 0.097. c 2 R 0.068. d 2 R 0.113. p < .05. p < .01. p < .001 (two-tailed).

This model assessed whether any effect of timing of sexual onset on later depressive symptomatology remained after the inclusion of the biopsychosocial predictors. When Wave 3 depressive symptomatology was simultaneously regressed on the biopsychosocial variables and age of rst coitus, the predictive effect of age of rst sexual intercourse on depressive symptomatology was no longer signicant (see Table 5, Model 3). Wave 1 depressive symptomatology remained a signicant predictor of Wave 3 depressive symptomatology (t 14.58, p < .001), indicating a signicant degree of stability over time (see Table 5, Model 4). Thus, the predictive effect of age of rst coitus on both Wave 3 depressive symptomatology and changes over time in depressive symptomatology disappeared when biopsychosocial predictors were included in the nal model.

Discussion Results from this study indicate that age of rst sexual intercourse was predictive of both subsequent depressive symptoms, as well as changes in levels of depressive symptoms. However, after controlling for biological, psychological, and sociological factors, age of rst coitus did not predict higher levels of subsequent depressive symptomatology, nor did it predict a change in depressive symptomatology. Recall that depressive symptomatology and early age of sexual intercourse were dependent variables. Therefore, it appears that

the relationship between age of rst sexual intercourse and depressive symptomatology may be spurious in that both are predicted by other variables. These results support Sabias (2006) conclusion that coitus appears to be an observable indicator of depression (p. 821) that is, one may be able to use either early coitus or early depressive symptomatology as a marker for the subsequent occurrence of the other, but early coitus and psychological distress do not appear to occur as a result of the other. As such, one needs to address the process that may explain both the increased likelihood of early sexual onset and higher levels of depressive symptomatology. Using a biopsychosocial framework, we identied some of these other variables. However, although the ndings of this study were statistically signicant, little variance was explained in the nal models, leaving a large portion of the variance in depressive symptomatology and early intercourse unaccounted for. What might some of these other factors be? Previous research has implicated deciencies in monoamines (serotonin, norepinephrine, and dopamine; Duval et al., 1997; Kalia, 2005), an increase of a corticotrophin-releasing factor due to hyperactivity in the hypothalamic pituitaryadrenal axis (Duval et al., 1997; Nemeroff & Vale, 2005; Taylor, 2003), and an increase in glucocorticoid levels (Kalia, 2005; Taylor, 2003) in the neurobiology of depression. The developing brain of an adolescent undergoes periods of vulnerability during synaptic pruning when glucocorticoid levels, the 457

JAMIESON AND WADE

hippocampus, prefrontal cortex, and amygdala may be over-sensitive to environmental inuences that lead to the development of depression. It appears that during this time of vulnerability, neurobiological changes may occur that alter brain chemistry, leading to the unfolding of depression (Andersen & Teicher, 2008). Previous research has also provided some indication focusing on a teens current and previous social and familial environment. Research among adolescents has identied higher peer connectedness as a risk factor for depression and risk-taking behaviors, whereas higher family and school connectedness and attachment are protective factors (Wade, 2001; Wade & Pevalin, 2005). Research on adults has shown that childhood maltreatment (including psychological, physical, and sexual abuse), as well as household dysfunction, were predictive of adult depression, increased sexual partners, and unplanned pregnancies (Chapman et al., 2004; Dietz et al., 1999). Strengths Past empirical evidence has focused on the relationship between early age of coitus and depression over approximately 12 months. This study addressed the relationship between early coitus and psychological distress over a seven-year time period, providing results that are representative of the process through adolescence and into early adulthood. Moreover, it looked at the long-term mental health consequences. Granted, short-term effects on mental health resulting from earlier age of rst coitus may exist, but these are likely transitory. If these effects are not transitory, then they would be identied in a longer-term assessment. In addition, short-term effects of early coitus on depression are difcult to determine based on the previous literature because of the limitation of only having a one-year interval between data-collection cycles. Many of the adolescents, especially those in the earlier grades, would not have had sufcient time to begin to engage in sexual relations. For example, many of those in Grade 7 in Wave 1 would have been in Grade 8 in Wave 2. Although some did have their rst sexual intercourse by eighth grade, this was a minority of students. As such, the majority of students in the early grades were deleted in previous literature, leading to the second strength of this studyits ability to account for most cases of right-hand censoring due to age. In this study, we were able to account for the occurrence of rst sexual intercourse in teens over a longer time period by analyzing adolescents over seven years. The original Add Health survey included those in Grades 7 to 12 in 1994. Because those adolescents in seventh grade at Wave 1 would likely have completed high school at the time of the third wave, the early-grade students were given much greater time to have had their rst sexual experience over seven years. Moreover, as the onset of the rst episode of depression generally occurs in later adolescence or early adulthood 458

(Robins et al., 1991), they were also given an additional opportunity to manifest depressive symptoms. Past empirical research has not been able to address the issue of right-hand censoring due to its cross-sectional or one-year longitudinal design. Therefore, this study extends previous research capturing the majority of teens as they progress through their adolescent years, minimizing the effect of right-hand censoring. This was also the rst study to ground the relationship between age of rst coitus and depressive symptomatology within a theoretical framework to assess other factors that may inuence both depressive symptomatology and age of onset of sexual intercourse. Although previous research utilized a handful of biological and sociological factors, there was little rationale for the inclusion of various covariates and control variables. Using the biopsychosocial framework (Engel, 1977), we examined a range of factors within each dimension. Although there was still much unexplained variance remaining, this study provided a more comprehensive examination than past research, examining how these various dimensions affect the relationship between earlier coitus and psychological distress. Finally, this study explored the bidirectional relationship between depressive symptomatology and earlier age of rst sexual intercourse, or the question of which comes rst? Past literature provided contradictory evidence of a bidirectional relationship, generally exploring the research question in one direction. The focus on one direction at the neglect of the potential of any bidirectional relationship implies a certain bias in perspective that was not grounded in a theoretical framework. This bias may partly explain the contradictory ndings of previous studies, as they were unable to demonstrate whether earlier coitus predicted subsequent depressive symptomatology or whether depressive symptomatology predicted earlier age of rst sexual intercourse. This study suggests that other factors were likely a result of a spurious relationship between sexual onset and depressive symptomatology, as both were explained by biological (age, sex, race, and pubertal maturity), psychological (self-esteem and social support), and sociological determinants (SES and religious involvement). Limitations The Add Health is a nationally representative longitudinal study from the United States that began in 1994. Thus, this study provides a comprehensive analysis of the sexual activity and mental health of the adolescent U.S. population that began in 1994, sampling students from Grade 7 to Grade 12. However, as with any longitudinal cohort study, it is only generalizable to people of the age cohort originally sampled. As such, it cannot generalize to adolescents in school now. Without some process in place to continually replenish and follow up younger ages as they enter this life stage,

SEXUAL INTERCOURSE AND DEPRESSIVE SYMPTOMATOLOGY

as the sample cohort ages, any cohort study becomes limited to the originally sampled cohort. In addition, because this sample comes from the United States, it is unclear how these results would generalize to other countries. Finally, we examined a heterosexual population, due to the independent associations for same-sex attraction and depression. The relationships identied among heterosexual teens may or may not exist among homosexual, bisexual, or asexual adolescents. The total variance explained for both depression and age of sexual onset was small even with the inclusion of these additional variables. This is, in part, a consequence of having limited content to include. Specically, there is an underrepresentation of psychological assessment measures in the Add Health data. Additional psychological variables may improve the variance explained in the relationship between psychological distress and age of rst intercourse. However, although underrepresented, the psychological variables in this study did provide signicant results. There were also many biological factors predictive of depression and early sexual onset that we were unable to examine, as they were not assessed by the Add Health (for a further review, see Jamieson, 2008). However, even with the limited number of biological factors included, we demonstrated the absence of any directional link between early sexual intercourse and depressive symptoms. The explanation for the relation between early rst intercourse and depression may lie in some of the untested biological variables mentioned or other factors we were unable to examine. In addition, this study followed a biopsychosocial theoretical framework to examine early coitus and depressive symptomatology. Thus, the research questions were designed to clarify the possible bidirectional relationship between these two variables, and, as such, the additional relationships that exist between the biological, psychological, and sociological variables were not assessed. The biopsychosocial theory separates the interaction of biological, psychological, and sociological factors among themselves. Thus, we did not assess sociocultural constructions, such as ethnicity or gender identity. Rather, we assessed the primary factors that constitute these constructions. Finally, this study used eight items of the 20-item CESD Scale. Whereas Waves 1 and 2 contained 18 of the 20 items of the CESD, in an effort to add content to the survey in Wave 3, the original architects of the study reduced the number of items to measure psychological distress. Although this may reduce the utility of the CESD, a factor analysis of these eight items revealed a similar factor structure to previous analyses of the 20-item scale (Radloff, 1977; Roberts et al., 1990), leading to condence in our results. This study is an improvement on previous work that used only one item of the CESD Scale, which was dichotomized, to assess depression (Rector et al., 2003; Sabia, 2006).

Implications This study addressed some of the limitations in previous research that led to the contradictory ndings in the literature. As such, it improves on past research and lays a foundation from which this relationship can be further examined. Our ndings, along with Meiers (2007) research, suggest that age of rst sexual intercourse is not signicantly related to future mental distress, except as a marker of previous factors. Based on the results of this study, by changing sexual behavior, one is not going to prevent future mental distress. It is important to look past the concomitant outcome and understand the underlying processes and risk factors involved that predict both earlier sexual intercourse and depressive symptomatology. Interventions directed toward youth, based on the factors that inuence both early coitus and depressive symptomatology, in sexual health campaigns would be benecial for adolescents.

References
Alexander, C. S., Somereld, M. R., Ensminger, M. E., Johnson, K. E., & Kim, Y. J. (1993). Consistency of adolescents self-report of sexual behavior in a longitudinal study. Journal of Youth and Adolescence, 22, 455471. Andersen, S. L., & Teicher, M. H. (2008). Stress, sensitive periods and maturational events in adolescent depression. Trends in Neurosciences, 31, 183191. Anderson, A. B., Basilevsky, A., & Hum, D. P. J. (1983). Missing data. In P. H. Rossi, J. D. Wright & A. B. Anderson (Eds.), Handbook of survey research: Quantitative studies in social relations (pp. 415 494). New York: Academic. Bogaert, A. F. (2004). Asexuality: Prevalence and associated factors in a national probability sample. Journal of Sex Research, 41, 279287. Bogaert, A. F. (2006). Toward a conceptual understanding of asexuality. Review of General Psychology, 10, 241250. Bogaert, A. F. (2008). Asexuality. In J. T. Sears (Ed.), The Greenwood encyclopedia of love, courtship, & sexuality through history (Vol. 6, pp. 1820). Westport, CT: Greenwood. Centers for Disease Control and Prevention. (2007). Youth risk behavior survey. Washington, DC: Author. Retrieved from http:// apps.nccd.cdc.gov/youthonline Chantala, K., & Tabor, J. (1999). National Longitudinal Study of Adolescent Health: Strategies to perform a design-based analysis using the Add Health data. Chapel Hill, NC: Carolina Population Center. Chapman, D., Whiteld, C., Felitti, V., Dube, S., Edwards, V., & Anda, R. (2004). Adverse childhood experiences and the risk of depressive disorders in adulthood. Journal of Affective Disorders, 82, 217225. Dietz, P. M., Spitz, A. M., Anda, R. F., Williamson, D. F., McMahon, P. M., Santelli, J. S., et al. (1999). Unintended pregnancy among adult women exposed to abuse or household dysfunction during their childhood. Journal of the American Medical Association, 282, 13591364. Duval, F., Mokrani, M. C., Crocq, M. A., Balley, P., Diep, T. S., a, H., et al. (1997). Dopamine function and HPA axis activity Corre in psychotic depression. Biological Psychiatry, 42, 111S112S. Engel, G. L. (1977). The need for a new medical model: A challenge for biomedicine. Science, 196, 129136.

459

JAMIESON AND WADE Hallfors, D. D., Khatapoush, S., Kadushin, C., Watson, K., & Saxe, L. (2000). A comparison of paper vs. computer-assisted self interview for school alcohol, tobacco, and other drug surveys. Evaluation and Program Planning, 23, 149155. Hallfors, D. D., Waller, M. W., Bauer, D., Ford, C. A., & Halpern, C. T. (2005). Which comes rst in adolescenceSex and drugs or depression? American Journal of Preventative Medicine, 29, 163170. Hallfors, D. D., Waller, M. W., Ford, C. A., Halpern, C. T., Brodish, P. H., & Iritani, B. (2004). Adolescent depression and suicide risk: Association with sex and drug behavior. American Journal of Preventative Medicine, 27, 224231. Harris, K. M. (2009). The National Longitudinal Study of Adolescent Health (Add Health), Waves I & II, 19941996; Wave III, 20012002; Wave IV, 20072009 [Machine-readable data le and documentation]. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Harris, K. M., Halpern, C. T., Whitsel, E., Hussey, J., Tabor, J., Entzel, P., et al. (2009). The National Longitudinal Study of Adolescent Health: Research design [Online] Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Retrieved from http://www.cpc.unc.edu/projects/ addhealth/design Jamieson, L. K. (2008). The relationship between early age of rst sexual intercourse and vulnerability to depression among adolescents. Masters thesis, Brock University, St. Catharines, Ontario, Canada (Theses Canada, Library, and Archives, AMICUS No. 36451137, ISBN: 9780494465813). Kalia, M. (2005). Neurobiological basis of depression: An update. Metabolism Clinical and Experimental, 54, 2427. Kaltiala-Heino, R., Kosunen, E., & Rimpela, M. (2003). Pubertal timing, sexual behavior and self-reported depression in middle adolescence. Journal of Adolescence, 26, 531545. Kessler, R. C., & Greenberg, D. F. (1981). Linear panel analysis: Models of qualitative change. New York, NY: Academic. Lam, T. H., Stewart, S. M., Leung, G. M., Lee, P. W., Wong, J. P., Ho, L. M., et al. (2004). Depressive symptoms among Hong Kong adolescents: Relation to atypical sexual feelings and behaviors, gender dissatisfaction, pubertal timing, and family and peer relationships. Archives of Sexual Behavior, 33, 487496. Lehrer, J. A., Shrier, L. A., Gortmaker, S., & Buka, S. (2006). Depressive symptoms as a longitudinal predictor of sexual risk behaviors among US middle and high school students. Pediatrics, 118, 189200. Meier, A. M. (2007). Adolescent rst sex and subsequent mental health. American Journal of Sociology, 112, 18111847. Molnar, B. E., Buka, S. L., & Kessler, R. C. (2001). Child sexual abuse and subsequent psychopathology: Results from the national comorbidity survey. American Journal of Public Health, 91, 753760. Nemeroff, C. B., & Vale, W. W. (2005). The neurobiology of depression: Inroads to treatment and new drug discovery. Journal of Clinical Psychiatry, 66, 513. Radloff, L. S. (1977). The CESD Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385401. Radloff, L. S. (1991). The use of the Center for Epidemiologic Studies Depression Scale in adolescents and young adults. Journal of Youth and Adolescence, 20, 149166. Rector, R. E., Johnson, K. A., & Noyes, L. R. (2003). Sexually active teenagers are more likely to be depressed and to attempt suicide. A report of the Heritage Center for Data Analysis. Washington, DC: The Heritage Foundation. Roberts, R. E., Andrews, J. A., Lewinsohn, P. M., & Hops, H. (1990). Assessment of depression in adolescents using the Center for Epidemiologic Studies Depression Scale. Psychological Assessment, 2, 122128. Robins, L. N., Locke, B. Z., & Regier, D. A. (1991). An overview of psychiatric disorders in America. In L. N. Robins & D. A. Regier (Eds.), Psychiatric disorders in America: The epidemiologic catchment area study (pp. 328366). New York, NY: Free Press. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Sabia, J. J. (2006). Does early adolescent sex cause depressive symptoms? Journal of Policy Analysis and Management, 25, 803825. Scoureld, J., Rice, F., Thapar, A., Harold, G. T., Martin, N., & McGufn, P. (2003). Depressive symptoms in children and adolescents: Changing aetiological inuences with development. Journal of Child Psychology and Psychiatry, 44, 968976. Smith, L. H., Guthrie, B. J., & Oakley, D. J. (2005). Studying adolescent male sexuality: Where are we? Journal of Youth and Adolescence, 34, 361377. Supple, A. J., Aquilino, W. S., & Wright, D. L. (1999). Collecting sensitive self-report data with laptop computers: Impact on the response tendencies of adolescents in a home interview. Journal of Research on Adolescence, 9, 467488. Taylor, S. E. (2003). Health psychology. (5th ed.). New York: McGraw-Hill. Tubman, J., Windle, M., & Windle, R. (1996). The onset and cross-temporal patterning of sexual intercourse in middle adolescence: Prospective relations with behavioral and emotional problems. Child Development, 67, 327343. Turner, C. F., Ku, L., Rogers, S. M., Lindberg, L. D., Pleck, J. H., & Sonenstein, F. L. (1998). Adolescent sexual behavior, drug use, and violence: Increased reporting with computer survey technology. Science, 280, 867873. Wade, T. J. (2001). Delinquency and health among adolescents: Multiple outcomes of a similar social and structural process. International Journal of Law and Psychiatry, 24, 447467. Wade, T. J., Cairney, J., & Pevalin, D. J. (2002). Emergence of gender differences in depression during adolescence: National panel results from three countries. Journal of the American Academy of Child and Adolescent Psychiatry, 41, 190198. Wade, T. J., & Pevalin, D. J. (2005). Adolescent delinquency and health. Canadian Journal of Criminology and Criminal Justice, 47, 619654.

Appendix1 Eight-Item Center for Epidemiological StudiesDepression Scale How often was each of the following true during the last week? 1. You were bothered by things that usually dont bother you. 2. You felt that you could not shake off the blues, even with help from your family and friends. 3. You felt that you were just as good as other people. 4. You had trouble keeping your mind on what you were doing. 5. You felt depressed. 6. You enjoyed life. 7. You felt sad. 8. You felt that people disliked you.
1

Reverse-coded items are indicated by an asterisk.

460

Copyright of Journal of Sex Research is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.