The Role of Sexual Behavior in the Identification Process of Gay and Bisexual Males.

Sexual behavior plays a significant role in the development of sexual-minority (gay and bisexual) males. Research spanning the last three decades illustrates that sexual-minority males exhibit greater sexual freedom--engaging in more sex with partners (Blumstein & Schwartz, 1983), meeting partners in highly sexualized environments (Blumstein & Schwartz, 1983), approving of sex without love (Klinkenberg & Rose, 1994; Lever, 1994; Tripp, 1975), reporting more sex partners (Blumstein & Schwartz, 1983; Lever, 1994), and developing sexually nonexclusive romantic relationships (Blumstein & Schwartz, 1983; Kurdek, 1989; McWhirter & Mattison, 1984)--than their heterosexual and lesbian counterparts.

Extant research suggests that sexual behavior facilitates the development of close relationships and the garnering of friends (Klinkenberg & Rose, 1993; Nardi, 1992). But sexual freedom and its significance have not been embraced by all gay and bisexual males. Many males do not engage in sex with numerous partners, and many same-sex male couples report a lifetime of sexual exclusivity (Bryant & Demian, 1994; Deenen, Gijs, & van Nearssen, 1994; Kurdek, 1989; Savin-Williams, 1998). More specifically, one study found that younger cohorts of sexual-minority males place less emphasis on sexual behavior and the sexual aspect of close relationships than do older cohorts of males (Deenen et al., 1994). What role does sexual activity play in the development of sexual-minority males? Can a model of sexual identity development account for the diversity found within these populations? Is age cohort the best predictor of this variance?

The current research addresses these questions by testing whether the timing of first sexual encounter is common across age cohort, and whether the timing of this event better predicts levels of adjustment to sexual identity and lifetime relationship involvement than age cohort alone.

PREVIOUS RESEARCH

Coming-Out Models

Sexual identity development models, also known as coming-out models, suggest that sexual-minority males develop same-sex attractions during childhood, engage in sex with other males during early adolescence, and first identify themselves as gay or bisexual during late adolescence or early adulthood (Cass, 1979; Coleman, 1982; Troiden, 1989). Empirical investigations into this process have generally supported this progression from sexual exploration with same-sex peers to self-labeling for White sexual-minority males (D'Augelli, 1991; Herdt & Boxer, 1993; Savin-Williams, 1990, 1998; Sears, 1991; Weinberg, 1978). Thus, sexual identity models predict that sexual behavior aids in the understanding and identification of attractions and personal identity. It is important to note that these writers do not infer that sexual behavior causes individuals to identify as gay; rather, sexual behavior confirms their suspicions that they are not heterosexual and helps construe their attractions and behavior into a personal identity. Taken collectively, these findings suggest that same-sex sexual behavior may be a catalyst for sexual identity formation.

These models do not, however, describe the large number of sexual-minority males who do not use sexual behavior to arrive at a gay identity. More specifically, coming-out models do not outline a sequence in which sexual-minority males label their sexual identity before engaging in sex with other males. The absence of alternate pathways to a sexual identity may be due to the common coming-out experiences of older males, who comprised most research samples used to investigate the coming-out process. Some research, however, has shown that a significant number of males report labeling their sexual identity before engaging in sexual behavior with other males (D'Augelli, 1991; Savin-Williams, 1998).

Age Cohort and Sexual Identity Development

One way of understanding within-group diversity among sexual-minority males is age cohort. The influence of age cohort on sexual identity development has received moderate attention from psychologists (Boxer & Cohler, 1989). In the 1970s and early 1980s, psychologists focused on the process by which gay males established a positive identity. This research frequently used samples of self-identified White males living in urban areas because these populations were the most readily available for and willing to participate in social science research. Consequently, much of the current literature relies on the experiences of these nonrepresentative groups.

Limited social resources existed for these gay males, beyond the bars and bathhouses in which many socialized. Few males were able to develop support networks beyond those they developed in sexualized contexts. Thus, many sexual-minority males from this era labeled and disclosed their sexual identity in highly sexualized contexts, and equated their sexual identity with their sexual behavior. The resultant models of research on these populations reflected the nearly universal experience of establishing a sexual identity only after years of denial, questioning, and same-sex sexual behavior. In fact, Weinberg (1978) found that only 2 of the 30 males he interviewed had reported labeling themselves as gay before ever engaging in sex with other males. The remaining 28 males identified themselves as gay after years of same-sex sexual behavior.

Because of their greater awareness and the social resources available to them in recent years, younger cohorts of sexual minorities may be better able to adjust to their same-sex attractions and may achieve developmental milestones at earlier ages. Research has generally supported these predictions (Savin-Williams, 1996). Savin-Williams (1998) reported that the coming-out process among young males occurs over fewer years than did that of older males, and that the younger males achieve most developmental events at significantly earlier ages. In addition, recent cohorts of gay and bisexual youths have come of age and identified their same-sex attractions in a distinctly less homophobic environment than did their older counterparts. In one report D'Augelli (1991) discovered that half of his sample labeled their sexual identity before engaging in sex, a distinct shift from Weinberg's (1978) sample 20 years earlier, in which the majority of males engaged in sex before labeling as gay or bisexual.

Older cohorts of sexual-minority males encountered greater barriers not only in adopting a sexual identity, but also in developing close relationships, which may also vary by age cohort. Most notably, recent cohorts of sexual-minority males reported fewer sexual encounters and greater romantic involvement (Diamond, Savin-Williams, & Dube, 1999). The decreased stigma on same-sex relationships and increased access to other sexual-minority youths in recent years may account for a potential cohort effect. Similarly, cohort effects have emerged in the value males place on close relationships. Deenan et al. (1994) discovered that older males reported more lifetime sex partners--a difference resulting from either age cohort or general development--and place less value on the emotional aspects of their relationships than younger males. Deenan's research did not, however, control for age in their analyses of lifetime sex partners. In short, these findings suggest that cohort may be a potentially powerful predictor of sexual identity development and relationship involvement.

Adjustment to Sexual Identity

Coming-out models invariably discuss two implications associated with adopting a gay sexual identity: adjusting to that identity and becoming involved with other sexual minorities. Adjustment to sexual identity most often includes disclosing the new identity to others and developing a positive attitude about one's sexual identity (or equally, achieving a low level of internalized homophobia). Individuals who grow up in greater homophobic environments should exhibit lower levels of adjustment (decreased disclosure of one's sexual identity to others and increased levels of internalized homophobia). There is reason to suspect that younger cohorts of sexual-minority males experience greater ease in adjusting to their sexual identity because of the shift in acceptance of sexual minorities in recent years. More specifically, sexual minorities that encounter positive images can not only develop a sense of belonging and self-acceptance, but also acquire the concepts and language necessary to label their attractions and behavior.

Disclosure. According to extant coming-out models, the disclosure process involves telling others that one is not heterosexual. Disclosure to others often represents the hallmark of the coming-out process, which occurs primarily after self-labeling. Savin-Williams (1998) found that most young gay males first disclosed to another person years after self-labeling, and through disclosure many males are able to develop a positive view toward their sexuality.

Internalized homophobia. Sexual minorities possess attitudes and beliefs about their same-sex attractions that reflect the views of the greater culture. Most often, these attitudes represent the stigma and negativity (homophobia) against individuals with same-sex attractions and behavior. As a result, sexual minorities learn to believe that same-sex sexuality is inferior and worthy of shame; collectively, these beliefs are commonly known as internalized homophobia (Nungesser, 1983; Shidlo, 1994). Although sexual minorities face difficulties in reducing levels of internalized homophobia, coming-out models and the extant research demonstrate a negative association between coming out and internalized homophobia--those who label and disclose their sexual identity to more individuals show lower levels of internalized homophobia (Savin-Williams, 1990; Shidlo, 1994).

Relationship Involvement

Heterosexual involvement. Although gay males report primarily same-sex attractions and behavior, many sexual-minority males also have heterosexual romantic and sexual relationships (D'Augelli, 1991; Savin-Williams, 1998; Sears, 1991). Reasons for such involvement include having same-sex and heterosexual attractions, experimenting, and denying same-sex attractions (Diamond et al., 1999). For instance, a young male questioning his sexual identity may develop a romantic and sexual relationship with a female peer to test whether he can establish and enjoy heterosexual relationships. According to this approach, males who engage in heterosexual relationships, but do not identify as bisexual, do so in route to identifying as exclusively gay.

Coming-out models predict that males who do not identify or feel comfortable with a gay identity are more likely to become sexually or romantically involved with females. More generally, heterosexual involvement among gay males is a function of social context (opportunities for developing a gay identity and same-sex relationships). Because recent cohorts of youths question their same-sex attractions earlier in their developmental trajectory, in more conducive environments, and with greater opportunity for same-sex involvement, these youths should be less likely than older males to report heterosexual involvement.

Same-sex involvement. Romantic and sexual relationships with other males carry greater implications than heterosexual relationships because society stigmatizes male/male intimacy (McCreary, 1994; Pleck, 1981). Consequently, many sexual-minority males either eschew such relationships, participate in them secretively, or develop them only when they feel comfortable with their same-sex attractions (Diamond et al., 1999). Because same-sex romantic relationships are an important aspect of gay males' lives, coming-out models theorize that gay males do not develop these relationships until the final stages of sexual identity development (Cass, 1979; Diamond et al., 1999; Troiden, 1989). Research in this area demonstrates that many males report this progression, from identification as gay or bisexual to same-sex romantic involvement (Savin-Williams, 1996).

Same-sex sexual relationships, however, carry fewer implications for sexual identity. The strongest evidence to support this comes from research showing that males engage in same-sex sexual behavior long before identifying as gay or bisexual (Cass, 1984; Savin-Williams, 1998; Weinberg, 1978). Some researchers suggest that sexual minorities engage in same-sex sexual relationships because these relationships are not subject to public scrutiny and involve minimal acknowledgment beyond the two people involved (Diamond et al., 1999). Further evidence, derived from cross-cultural research, demonstrates that same-sex sexual behavior may have little to do with sexual identity, and may be independent of either social context or adjustment to sexual identity for many males (Dube, Savin-Williams, & Diamond, in press; Manalansan, 1996). For instance, many Black and Latino males engage in same-sex sexual relationships without ever identifying as gay or bisexual. Finally, a review of the empirical research on the coming-out process illustrates that males over the last three decades report first same-sex sexual experience around the age of 13 years, several years before applying the concept of a gay identity to themselves (for review see Savin-Williams, 1996). This robust finding demonstrates that even when males do not understand their same-sex attractions in the context of an integrated identity, they act sexually on these attractions. More importantly, these males may understand their same-sex attractions in terms of their behavior.

The Current Study

The developmental sequence put forth by established coming-out models can be termed sex-centered, in which males use same-sex sexual behavior to understand and label their same-sex attractions. As predicted by these early theories of sexual-identity development, males who develop through this sequence rely on sex to form their personal identity and sense of being gay (Weinberg, 1978). Males who engage in sex with other males prior to self-labeling thus develop through a sex-centered sequence. Males who follow an identity-centered sequence (i.e., identifying their sexual identity before engaging in sex with other males) may not view sex as the defining characteristic of their identity. For these males, contact with other sexual minorities, reading literature on sexuality, cognitive questioning, or fantasy during masturbation may serve as the catalyst for labeling their sexual identity; sexual behavior with other males would play a lesser role in the identification process of identity-centered males.

The current research tests how these two sequences (sex-centered and identity-centered) predict later adjustment to sexual identity and relationship involvement. As such, much of the research was exploratory in nature. Specific hypotheses, however, were formulated. The proportion of males that develop through each developmental sequence should be predicted by age cohort. A greater proportion of older males should report developing through a sex-centered sequence, whereas younger males, who have had greater access to identity labels and support resources, are more likely to proceed through an identity-centered sequence. Sequence differences are also expected to emerge in three areas: adjustment to sexual identity (i.e., disclosure, internalized homophobia), heterosexual relationship involvement, and same-sex sexual involvement.

Males who follow the sex-centered sequence construe their sexual identity only after engaging in sex with another male. The effect of this process is twofold. First, these males may experience a disjuncture between having sex and being gay. Often, males who follow a sex-centered sequence engage in sex with males for years without ever labeling themselves or their attractions as gay (D'Augelli, 1991; Savin-Williams, 1996; Weinberg, 1978). During this period these males, according to coming-out models, deny their attractions, question their sexuality, and struggle with an identity that integrates their attractions and behavior. If these males experience difficulty adopting a positive gay identity, then they should report relatively high levels of internalized homophobia and low levels of disclosure.

Second, sex-centered males might engage in different patterns of relationships as a result of their sexual identity. Coming-out models predict that many sexual-minority males engage in romantic and sexual relationships with females either because they have heterosexual attractions or because they hope to disguise their true sexuality by dating heterosexually. Among sex-centered males, who are predicted to experience a protracted questioning period, relationships with females may be more common.

Because sex-centered males identify themselves as gay once they are sexually active, sexual behavior may take on heightened importance. Sex-centered males are likely to perceive sex with other men as a primary aspect of their sexual identity. Consequently, these males should report having had many male sex partners throughout their life. Little research has been conducted on same-sex romantic relationships and the role of these relationships in the coming-out process. As such, no formal hypotheses are forwarded with regard to sequence differences in same-sex romantic involvement.

Identity-centered males label themselves as gay without ever engaging in sex with other males. The concept of a sexual identity may take on a distinct meaning among these males. That is, the sexual component of a gay identity may take a back seat to cognitive, political, emotional, and interpersonal aspects. As a result, these males are likely to report fewer lifetime sex partners than sex-centered males, even after controlling for potential age differences. Furthermore, identity-centered males label themselves early and after cognitively questioning their attractions. As a result, they may be better able to adjust to their sexual identity (e.g., high rates of disclosure and low rates of internalized homophobia). Similarly, identity-centered males should report lower rates of heterosexual involvement than sex-centered males because they identify and label their sexual identity earlier.

METHODS

Participants

One hundred sixty-six gay and bisexual males between the ages of 16 and 39 years completed questionnaires assessing measures of sexual-identity development and relationship involvement. Participants were categorized into three age cohorts. Fifty-six males between 16 and 21 years of age comprised the youngest group (cohort 1). The middle group contained 68 males between the ages of 22 and 29 years (cohort 2). Forty-two males between the ages of 30 and 39 were placed in the oldest group (cohort 3). Participants were recruited from community and college support groups, e-mail listservs, and word of mouth to participate in a study about relationships among males. Table 1 illustrates the sample distribution by recruitment venue. Inclusion criteria were (a) between the ages 16 to 40 years and (b) sexual or romantic attractions to other males. Three hundred questionnaires were distributed and 166 were received, for a response rate of 55%. The moderate total response rate may be accounted for by the unique and diverse recruitment efforts to achieve the most representative sample. To obtain a sample diverse in age, ethnicity, and geographical area, a high response rate is an unavoidable sacrifice. Moreover, questionnaires were completely anonymous to maximize the comfort of respondents; thus, follow-up contact to encourage response could not be established. This strategy was particularly important because ethnic sexual-minority males may be less likely to identify as gay or bisexual, or to volunteer for social science research. To achieve sufficient numbers of ethnic-minority males, anonymity was a necessity.

Table 1. Sample Demographics by Age Cohort

Demographic                        Cohort 1   Cohort 2   Cohort 3

Recruitment venue
  College group                      21%        10%        13%
  Relationship workshop              30%         2%         3%
  Email groups                        7%        22%        18%
  Community centers and hangouts     31%        48%        51%
  Friendship networks                11%        18%        15%

Childhood residence
  Urban                              23%        24%        17%
  Suburban                           45%        50%        37%
  Small town                         21%        15%        24%
  Rural                              11%        11%        22%

Current residence
  Urban                              39%        66%        64%
  Suburban                           32%        19%        14%
  Small town                         22%        10%        17%
  Rural                               7%         5%         5%

Ethnicity
  White                              54%        56%        60%
  Black                              14%         8%        17%
  Latino                             27%         8%         8%
  Asian American                      5%        28%        15%

Males from a variety of backgrounds comprised the final sample, which represents a considerable strength of the current research. Sequence groups did not differ in levels of education. Most males reported at least two years of college. Tests were not conducted to determine whether age/cohort groups differed in levels of education because many of the younger respondents were in high school at the time of the assessment.(1) Because the geographic area (Table 1) in which these males grew up may have influenced how they developed their sexual identity, it was necessary to test whether the age cohorts or sequence groups varied in where they grew up. No associations were found between either age cohort or developmental sequence and where the males grew up.

As shown in Table 1, most respondents currently live in an urban area. Maximum likelihood logistic regression models were fit to the data to test for sequence differences in current residence. Current location, the dependent variable, was coded as either urban or nonurban (suburban, small town, or rural). Sequence was entered as the independent variable, and age was included to control for potential age effects. Older males had greater opportunity to move than younger males, many of whom were in high school at the time of the study. The final model included significant terms for both age (p = .04) and sequence (p = .03), [chi square] (2, N= 143) = 11.15, p = .005. Results suggest that age was positively associated with moving to urban areas. More specifically, the likelihood that someone currently resided in an urban area increased with age. More importantly, sex-centered males were 1.5 times more likely than identity-centered males to reside in urban areas, even after controlling for age effects.

Recent theoretical writings contend that ethnicity is a significant context for sexual-identity development and relationship development (Chan, 1995; Dube & Savin-Williams, 1999; Dube et al., in press; Greene, 1997). As such, recruitment strategies were used to maximize the diversity within sexual-minority male populations. The youngest cohort contained slightly more Latino males than the older cohorts; the middle cohort group contained slightly more Asian American males. Although differences emerged in the ethnic composition of each age/cohort group, each group contained a sufficient number of each cohort group and was comprised of at least 40% ethnic minority males. The diversity with regard to age, geographic location, recruitment venue, and ethnicity are considerable strengths of the current research.

Measures

Participants were asked to provide the ages at which they experienced each milestone in the development of a sexual identity (age of first awareness of same-sex attractions, age of first same-sex sexual encounter, age of identification of self as gay, age of first disclosure to other, age of first same-sex romantic relationship). From these ages, primary sequence was coded as either sex-centered (label sexual identity after first sexual experience with a male) or identity-centered (identification of sexual identity before first sexual encounter with a male). Individuals who labeled their attractions and first engaged in sex with a male at the same age were not grouped because of the temporal proximity of these events. A one-item Kinsey-type scale was administered to measure the participants' current sexual identity label. Participants were asked to select one of seven identity labels that best described their current attractions and behavior, ranging from completely heterosexual to completely gay.

Adjustment to sexual identity was assessed by levels of disclosure and internalized homophobia. To measure disclosure levels participants provided the ages at which they told others about their same-sex attractions. Participants were given a list of 18 individuals ranging from parents to teachers and members of the clergy. Items on the disclosure scale were computed into one variable based on number and status of persons told. Disclosure scores ranged from 1 (no individuals told) to 7 (all known individuals told) and were based on Guttman-like scoring. A higher score (6 or 7) was given to respondents who disclosed to their parents and close peers, if these individuals were present in the youth's life. Lower scores (1-3) reflect more limited disclosure; for example, to sexual-minority peers or distant individuals, such as a teacher or therapist. More weight was given to parents and close friends because disclosure to these individuals is more important and entails greater risk for youths (D'Augelli, Hershberger, & Pilkington, 1998; Hershberger, Pilkington, & D'Augelli, 1998)

Internalized homophobia was measured by a revised version of Nungesser's Homosexual Attitudes Inventory (NHAI), consisting of three subscales: homophobia about self (11 items), others (11 items), and disclosure (14 items). Internalized homophobia about self reflects attitudes toward the fact of one's homosexuality (e.g., "Whenever I think a lot about being gay/bisexual, I feel depressed"); homophobia about others reflects attitudes toward homosexuality in general and toward other gay males (e.g., "Gay males are overly promiscuous"); and homophobia about disclosure reflects attitudes toward the fact of one's sexual identity being known by others (e.g., "I would not mind if my neighbors knew that I was gay/bisexual"). Several items were changed or deleted to reflect updated attitudes toward being gay, as the original instrument was developed nearly twenty years ago. Participants were given the 36-item questionnaire and asked to respond to each item using a 5-point likert-type scale, from strongly disagree to strongly agree. Higher scores on each scale reflect higher levels of internalized homophobia.

The reliability coefficient from Nungesser's original sample (Cronbach's alpha) for the total scale is .95 and for the three scales the coefficients are .88 (self), .67 (other), and .93 (disclosure). Within the current sample the overall reliability (Cronbach's alpha) reached .92. Additional correlation analyses indicated a significant level of construct validity, as well as an intimation that internalized homophobia decreases as an individual develops a stable sexual identity (Nungesser, 1983; Shidlo, 1994). This conclusion, however, is limited by the cross-sectional design of Nungesser's study. This measure of internalized homophobia is integral to the factor of sexual identity development because a person who has a positive sexual identity will be less likely to possess negative attitudes toward his homosexuality, or toward homosexuality in general (Nungesser; for review see Shidlo). It has been theorized that internalized homophobia decreases over time; this process parallels the identification process, in which internalized homophobia decreases around the same time males label and disclose their sexual identity. It can be argued, however, that internalized homophobia regarding oneself is conceptually distinct from internalized homophobia regarding other males. More specifically, identification and disclosure of a gay identity is more closely related to a decrease in the self and disclosure subscales of this measure of internalized homophobia. This process may not, however, result in a decrease in homophobia toward other males. To acknowledge these potential differences, analyses testing for age and sequence effects on internalized homophobia will treat the three subscales as distinct concepts.

Relationship involvement was determined by questions probing relationship histories. These items required respondents to report the number of same-sex and heterosexual romantic and sexual relationships and the ages at which first same-sex and heterosexual relationships occurred.

Procedures

Questionnaires were distributed by the primary investigator to community and school groups, through friends, and by request from those who saw public postings about the study. Additionally, a workshop on developing romantic relationships was designed to answer youths' questions about developing same-sex relationships. In these groups, youths completed the questionnaires and then participated in a discussion of relationships. Workshops were conducted at community and regional youth centers and were not announced as dealing with relationships. Thus, respondents from the workshops should not differ with regard to relationship interest or experience. Before completing the questionnaire, all participants were told that the study concerns males and their relationships and that their participation was completely voluntary.

RESULTS

Developmental Sequence and Age Cohort

Fifty-eight percent of males developed through the sex-centered sequence. Before any analyses were conducted, tests were conducted to determine whether the sequence groups varied in their Kinsey ratings and their age/cohort status. A Mann-Whitney U test demonstrated that the groups did not differ in their Kinsey rating. Figure 1 illustrates the proportion of males developing through these sequences by age/cohort. A chi-square test of association detected significant age/cohort differences in these sequences, [chi square] (2, N = 144) = 6.25, p = .04. As predicted, most males in the older (69%) and middle (62%) cohorts developed through a sex-centered sequence; far fewer males (44%) from the younger cohort proceeded through this sequence.

[Figure 1 ILLUSTRATION OMITTED]

The temporal period between age of first sex with a male and age of self-identification was measured by differences scores. Among males who followed a sex-centered sequence, an average age of 3.5 years (SD = 4.4) passed between first sex with a male and labeling. ANOVA models were fit to the data to test whether difference scores varied by age cohort. Cohort did not contribute significantly to the final model, F (2, 80) [is less than] .1. A similar analysis was conducted among males who followed an identity-centered sequence. Among these males, an average of 3.7 years (SD = 3.5) passed between age of self-identification and first same-sex sexual encounter. Age cohort could not explain a significant proportion of the variance, F (2, 58) [is less than] 1. These findings suggest that although the sequence of identification and first sex varied by cohort, the temporal period between these two events remained constant across age cohort.

Adjustment to Sexual Identity

Forty-four percent of the participants disclosed their sexual identity to all friends, family members, and acquaintances (a score of 6 or 7 on the disclosure scale). The distribution of disclosure scores was nonnormal, as most individuals have disclosed to friends and family. Consequently, analysis of variance could not be conducted to test for sequence differences in disclosure patterns. A Mann-Whitney U rank sum test was used to determine whether the sex-centered and identity-centered groups differed in disclosure, and revealed that disclosure patterns did not differ across sequence groups. A similar analysis revealed that age cohorts did not vary in levels of disclosure. Contrary to prediction, males who developed through a sex-centered or an identity-centered sequence disclosed their sexual identity to most known individuals.

Most males also achieved low levels of internalized homophobia regarding their own sexuality and homosexuality in general. Overall means for each internalized homophobia subscale demonstrated that males in the current sample reported positive outlooks toward their sexual identity and homosexuality in general (self scale, mean = 22.0, SD = 8.23; other scale, mean = 22.4, SD = 5.9; and disclosure scale, mean = 30.6, SD = 11.3). Multiple linear regression analysis was used to predict levels of internalized homophobia toward self and others by sequence and age cohort. No model fit the data well when internalized homophobia toward either self or disclosure was used as the dependent variable, suggesting that males across cohort and sequence achieved low levels of internalized homophobia regarding one's own sexual identity. Furthermore, males across sequence and cohort held similar views toward disclosing their sexual identity to others. When internalized homophobia toward others was selected as the dependent variable, however, a model with cohort and sequence fit the data well, F (2, 140) = 3.8, p [is less than] .05. Age cohort did not contribute significantly to the model, but remained in the final model to control for age. As predicted, males who followed a sex-centered sequence reported higher levels of internalized homophobia (mean = 23.4, SD = 5.9) toward others than males who developed through an identity-centered sequence (mean = 20.6, SD = 5.41). The ranges of internalized homophobia scores ranged from 11 to 42 within the total sample, with higher scores representing higher levels of internalized homophobia.

Relationship Involvement

Fifty-four percent of the males reported having had a sexual relationship with a woman. The same percentage also had developed a heterosexual romantic relationship. Most males having sex with females also reported romantic relationships with females, [chi square] (1, N = 161) = 76.4, p [is less than] .00001. Thus, sexual and romantic involvement with females appears to be, in most cases, linked. Of the males who followed an identity-centered sequence, 44% reported having had either a sexual or romantic relationship with a woman, while 71% of males who followed a sex-centered sequence reported either type of relationship. Maximum likelihood logistic regression was used to predict heterosexual involvement from sequence, Kinsey rating, and age. The first test used heterosexual sexual involvement as the dependent variable and yielded a highly significant final model, [chi square] (8, N = 141) = 20.01, p [is less than] .006, which included nonsignificant terms for age and Kinsey rating, and a significant sequence term. As expected, males following a sex-centered sequence were 1.7 times more likely than males in the identity-centered group to report having sex with females. Moreover, this finding remained constant across age cohort. Contrary to prediction, males reporting lower Kinsey ratings (greater heterosexual attractions and behavior) were not more likely to report heterosexual sex, suggesting moderate independence of sexual behavior and Kinsey ratings (i.e., sexual identity).

An analysis was conducted to determine whether the sex-centered and identity-centered groups varied in timing of first sexual involvement with a male and with a female partner. Of the 68 males who reported having had sex with both, 57% reported that they had engaged in sex with a woman before engaging in sex with a male. A chi-square test of independence, [chi square] (1, N = 68) = 7.96, p [is less than] .005, determined that identity-centered males were significantly more likely to engage in sex with a woman (82%) before having sex with a male. Among the males in the sex-centered group, 46% reported heterosexual sex first. Thus, sex-centered males were more likely to have engaged in heterosexual sex after they had sex with males. Whether these men engaged in heterosexual sex after they identified as nonheterosexual remains largely unknown.

Rates of heterosexual romantic involvement were also tested for group differences. A maximum logistic regression model predicting heterosexual romantic relationships was significant, [chi square] (8, N = 142) = 30.62, p = .0001, with significant terms for Kinsey rating and sequence, controlling for age. With regard to romantic involvement with females, males who gave lower Kinsey ratings were more likely to report at least one heterosexual romantic relationship. Males who followed a sex-centered trajectory were almost twice as likely as males who followed an identity-centered sequence to report a heterosexual romantic relationship. Thus developmental sequence, not age cohort, best predicted heterosexual romantic and sexual involvement.

Males in the current sample reported an average of 3.5 (SD = 4.8) lifetime romantic relationships with males. Sex-centered males reported a mean of 4.0 (SD = 5.8) relationships; identity-centered males, an average of 3.0 (SD = 3.5). Analysis of variance tested the effects of sequence and age on same-sex romantic histories. As predicted, neither sequence nor age contributed to the final model, suggesting that males across groups did not vary in number of romantic relationships with males.

The distribution of past male sex partners was positively skewed. The median number of male sex partners for this sample was 11; the mean, 14 (SD = 4.4). The number of lifetime male sex partners was logged to normalize the overall distribution. Males in the sex-centered group reported an average of 19 (SD = 3.9) past sex partners; males in the identity-centered group, an average of 9 (SD = 4.7). Multiple linear regression testing for age cohort and sequence effects yielded significant results, F (2, 128) = 10.74, p [is less than] .0001. Both age cohort and sequence contributed significantly to the overall model. As predicted, older males and males who followed the sex-centered sequence reported higher numbers of male sex partners than either younger males or males in the identity-centered group. These findings, however, cannot determine whether sex-centered males engaged in high rates of sexual behavior before or after they identified. To determine whether sex-centered males had most sexual encounters before they identified, the rates of sexual involvement were correlated with difference scores between age of first sexual encounter and age of identification. No association was found for either the sex-centered or identity-centered group. These results suggest that sex-centered males may continue to pursue high numbers of sexual encounters with men after they identify.

DISCUSSION

Developmental sequence varied by age cohort; older males disproportionately developed through a sex-centered sequence whereas younger males proceeded primarily through an identity-centered sequence. This differential model of sexual-identity development predicted internalized homophobia toward other males, heterosexual involvement, and lifetime sexual involvement better than established coming-out models or age cohort alone. Sex-centered males evidenced higher levels of internalized homophobia toward other sexual-minority males. No group differences were found with regard to disclosure and internalized homophobia toward self. Finally, sex-centered males were more likely to report heterosexual sexual and romantic involvement and showed higher numbers of past male sexual partners.

Developmental Sequence and Cohort

Results demonstrated that developmental sequence was cohort-specific; a greater proportion of current sexual-minority youths identified before engaging in sex than previous cohorts of youths. The increased proportion of identity-centered trajectories among current sexual-minority youths may represent a significant shift in the ways males organize and establish their sexual identity. The recent and unabated emergence of positive gay images and role models may play a central role in this shift. Television sitcoms, talk shows, and mainstream movies portraying positive, well-adjusted gay and lesbian individuals have infiltrated most regions of the United States. Consequently, youths with same-sex attractions may be encouraged to self-label early, prior to having experimented sexually. This process is unique to recent cohorts of gay youths, as positive images of sexual minorities were virtually nonexistent over a decade ago.

Furthermore, whereas sexual-minority males growing up before 1990 were likely to engage in sex with other males before self-labeling, sexual-minority youths growing up in the 1990s may seek forms of experimentation that do not involve sexual behavior, such as developing friends on the internet, reading gay-affirming materials, or attending community support groups. Moreover, fear over HIV infection may lead these youths to eschew early sexual involvement.

It is noteworthy, however, that diversity exists within each cohort group. A significant number of the older males developed through an identity-centered sequence, and many current youths report engaging in sex with male peers before self-labeling. Cultural values or opportunities for sexual partners may make a significant contribution to the observed within-group diversity. More specifically, youths who grow up in an environment promoting abstinence or rigid gender roles may be disproportionately likely to develop through either a sex-centered or identity-centered sequence.

Adjustment to Sexual Identity

Most males disclosed their sexual identity to friends and family. Neither age cohort nor sequence predicted the extent of disclosure. In addition, most males achieved a positive outlook regarding their sexual identity, as evidenced by the low levels of internalized homophobia. Although the sex-centered and identity-centered groups did not differ in overall levels of internalized homophobia about themselves, sex-centered males reported higher levels of internalized homophobia about others than identity-centered males. Perhaps these males, many of whom have had frequent sex without identifying as gay or bisexual, project their negative feelings about their same-sex attractions onto self-identified males with whom they have sex. Identity-centered males, on the other hand, might be more established in their sexuality at the time of their first sexual encounter.

Relationship Involvement

Heterosexual involvement. Approximately half of the current sample reported at least one heterosexual relationship. Those who reported heterosexual sexual relationships were also likely to report heterosexual romantic relationships. Males who reported lower Kinsey scores (more heterosexual scores) were more likely to report heterosexual romantic involvement. This finding suggests that the males in the current sample integrated only their romantic experiences into their overall sexual identity. The same association was not found between Kinsey scores and sexual involvement. Perhaps these men were able to justify engaging in heterosexual sex as experimentation or yielding to peer pressure. Thus, the behavior would have little bearing on their overall sexual identity label. Heterosexual romantic involvement, coupled with sexual behavior, may be more difficult to justify in light of sexual identity ratings. On the other hand, males with same-sex and heterosexual attractions may be motivated to develop sexual and romantic relationships with women. Future research should explore the motivations for and importance of romantic and sexual involvement with women among these groups of men.

Findings that most sexual-minority males engaged in sex with both male and female peers are consistent with earlier research (Rosario, Meyer-Bahlburg, Hunter, & Exner, 1997). These results demonstrate, however, that males who follow a sex-centered sequence were more likely than identity-centered males to report heterosexual relationships. Most often, these relationships occurred prior to self-labeling, suggesting that heterosexual involvement provided a context in which to explore and confirm same-sex attractions and sexual identity. Among identity-centered males, heterosexual relationship involvement occurred less frequently either because males without heterosexual attractions were able to label early in their development, or because males who labeled early understood and accepted their same-sex attractions without needing to experiment with heterosexual relationships. When identity-centered males reported heterosexual sexual involvement, it most often occurred before same-sex sexual involvement. Identity-centered men who reported heterosexual sex most likely had heterosexual attractions, or used these experiences to negotiate their sexual identity. The differences in heterosexual sex between sex-centered and identity-centered males are most likely not due to greater heterosexual attractions among sex-centered males, because Kinsey scores did not vary across sequence groups. More likely, sex-centered males used heterosexual sex to deny same-sex attractions, or to justify having sex with males without labeling themselves gay.

Same-sex romantic involvement. Most males reported at least one same-sex romantic involvement. On average, males developed these relationships during early adulthood, after labeling and disclosing their sexual identity. Rates of past romantic involvement did not vary by identity sequence or age cohort. These null findings suggest that sexual-minority males who identify and disclose their sexual identity establish romantic relationships with other males regardless of social context or developmental sequence. More importantly, however, if gay youths who are currently in the second decade of life report similar rates of same-sex romantic relationships as males currently in their late thirties, then it is likely that the younger males will develop more romantic relationships throughout their lifecourse. It is important to note, however, that the data only suggest this possible association; no causal relationship exists.

Same-sex sexual involvement. Rates of sexual involvement were more variable than rates of romantic involvement. Moreover, as expected given their greater opportunity by living more years, older males reported more male sex partners than younger males. Similarly, sex-centered males reported more sex partners than identity-centered males. The sequence difference presents a more complex process. Sex-centered males may report more sex partners because they engaged in high rates of sexual behavior prior to identification. Perhaps after they identified as gay or bisexual their rates of sexual involvement resembled the rates of identity-centered males. However, this is unlikely given that rates of sexual involvement were not associated with the length of time between the age of first sexual encounter and age of self-identification. More likely, sex-centered males reported higher rates of sexual behavior because they viewed sex as a central feature of their sexual identity. These males continue to engage in sex with many partners because they identify their sexual identity in the context of sexual behavior, and value sex with other males as an expression of their identity. Identity-centered males do not regard sexual behavior as a defining feature of their sexuality, perhaps because they arrived at their identity label after years of nonsexual questioning.

CONCLUSION

Age cohort represents a significant factor in the coming-out process of sexual-minority males, as younger and older sexual minority males showed different developmental trajectories and patterns of relationship involvement. Developmental sequence, however, provides a stronger and more parsimonious way of understanding developmental diversity among gay and bisexual males. More specifically, the timing of first same-sex sexual encounter may be the best available predictor of later adjustment to sexual identity and rates of same-sex and heterosexual sexual involvement. Developmental sequence may also help predict other aspects of sexual behavior, such as high-risk sexual behavior. Substantial research has investigated the role of anxiety and mental health as predictors of HIV risk (e.g., Rotheram-Borus, 1994; Rotheram-Borus, Reid, Rosario, & Kasen, 1995), yet few researchers have explored the link between sexual-identity development and risk-taking behavior. It is reasonable to expect that males who incorporate sexual behavior into their sexual identity, and thus engage in high rates of sexual behavior, may be at greater risk of HIV infection. Moreover, these findings have strong implications for adjustment to sexual identity. Individuals exposed to positive images early in their development may have an easier time adopting a positive sexual identity.

The current findings should, however, be taken with consideration of several limitations. Although not all males who participated in the current sample identified as exclusively gay, the current research relied primarily on the stories of self-identified sexual-minority males who selected to report about their sexual identity and close relationships. Moreover, the findings from this convenience sample can be generalized only to subpopulations of sexual-minority males, who have identified their same-sex attractions and were willing to participate in social science research. The sample used for the current research was not large enough to test for within-group differences beyond age cohort and developmental sequence, such as religion, socioeconomic status, or attitudes toward sexual behavior.

These self-reports may be vulnerable to several biases. First, research that relies on self-report is susceptible to retrospective reporting biases. Individuals who are asked to report on the ages at which specific events occurred may not be completely accurate. Whereas most self-report data invariably suffer from retrospective bias, some research suggests that gay men may accurately report early memories of sex-typed behavior and perhaps other developmental events (Bailey, Nothnagel, & Wolfe, 1995). Second, reports of sexual behavior may not be completely consistent over time (Stone, Catania, & Binson, 1999). Thus, the reports of older males in the current sample may reflect a larger distortion than those of younger males. The lack of interaction between age and sequence group in the prediction of lifetime sex partners, however, suggests a more complex effect if one assumes that these reports are vulnerable to distortion. More importantly, sequence differences emerged in the number of sex partners across all ages.

Future research must first confirm these findings before exploring how developmental sequence influences adjustment and relationship involvement. For example, this line of research could test whether learning theory can explain how sex-centered men continue to view sex as a central component of their sexual identity. Alternatively, research could explore whether models of ego-identity development predict the behavior of identity-centered and sex-centered males. Next, additional research should test whether this model of development can be generalized to broader populations of sexual-minority males: those who have not yet identified as gay or bisexual, those who identified their same-sex attractions later in life, and those who grew up in homophobic and nonhomophobic family environments. Understanding of sexual identity development could also greatly benefit from longitudinal research that follows youths as they first experience each developmental milestone. Finally, research should test for within-group differences in this process, such as gender, ethnicity, religion, gender role, and exposure to gay images.

(1) Participants under the age of 18 were included to maximize the diversity of the sample. Parental consent was not required because these youths may not have disclosed their sexual identity to their parents, and requiring consent would jeopardize the parent-child relationship. Age limits and forfeiture of parental consent were approved by IRB.

REFERENCES

Bailey, J. M., Nothnagel, J., & Wolfe, B. A. (1995). Retrospectively measured individual differences in childhood sex-typed behavior among gay men, a correspondence between self and maternal reports. Archives of Sexual Behavior, 24, 613-622.

Blumstein, P., & Schwartz, P. (1983). American couples: Money, work, sex. New York: Morrow.

Boxer, A., & Cohler, B. (1989). The life course of gay and lesbian youth: An immodest proposal for the study of lives. Journal of Homosexuality, 17, 317-355.

Bryant, A. S., & Demian (1994). Relationship characteristics of American gay and lesbian couples: Findings from a national survey. Journal of Gay & Lesbian Social Services, 1, 101-117.

Cass, V. (1979). Homosexual identity formation: A theoretical model. Journal of Homosexuality, 4, 219-235.

Cass, V. (1984). Homosexual identity: A concept in need of a definition. Journal of Homosexuality, 9, 105-126.

Chan, C. S. (1995). Issues of sexual identity in an ethnic minority: The case of Chinese American lesbians, gay males, and bisexual people. In A. R. D'Augelli & C. J. Patterson (Eds.), Lesbian, gay, and bisexual identities over the lifespan: Psychological perspectives (pp. 87-101). New York: Oxford University Press.

Coleman, E. (1982). Developmental stages of the coming-out process. In J. Gonsiorek (Ed.), Homosexuality and psychotherapy (pp. 31-44). New York: Hawthorne.

D'Augelli, A. R. (1991). Gay males in college: Identity processes and adaptations. Journal of College Student Development, 32, 140-146.

D'Augelli, A. R., Hershberger, S. L., & Pilkington, N. W. (1998). Lesbian, gay, and bisexual youth and their families: Disclosure of sexual orientation and its consequences. American Journal of Orthopsychiatry, 68, 361-371.

Deenen, A. A., Gijs, L., & van Nearssen, A. X. (1994). Intimacy and sexuality in gay male couples. Archives of Sexual Behavior; 23, 421-431.

Diamond, L. M., Savin-Williams, R. C., & Dube E. M. (1999). Sex, dating, passionate friendships, and romance: Intimate peer relations among lesbian, gay, and bisexual adolescents. In W. Furman, C. Feiring, & B. B. Brown (Eds.), Contemporary perspectives on adolescent romantic relationships (pp. 175-210). New York: Cambridge University Press.

Dube, E. M. & Savin-Williams, R. C. (1999). Sexual identity development among ethnic sexual-minority male youths. Developmental Psychology, 35, 1389-1398.

Dube E. M., Savin-Williams, R, C., & Diamond, L. D. (in press). Intimacy development among sexual minority youths. In A. R. D'Augelli & C. J. Patterson (Eds.), Lesbian, gay, and bisexual youth: Psychological perspectives. New York: Oxford University Press.

Greene, B. (Ed.). (1997). Ethnic minority lesbians and gay males: Mental health and treatment issues. Thousand Oaks, CA: Sage Publications.

Herdt, G., & Boxer, A. M. (1993). Children of Horizons: How gay and lesbian teens are leading a new way out of the closet. Boston: Beacon Press.

Hershberger, S. L, Pilkington, N. W, & D'Augelli, A. R. (1998). Predictors of suicide attempts among gay, lesbian, and bisexual youth. Journal of Adolescent Research, 12, 477-497.

Klinkenberg, D., & Rose, S. (1994). Dating scripts of gay males and lesbians. Journal of Homosexuality, 26, 23-35.

Kurdek, L. (1989). Relationship quality in gay and lesbian cohabiting couples: A 1-year follow-up study. Journal of Social and Personal Relationships, 6, 39-59.

Lever, J. (1994, August 23). Sexual revelations. The Advocate, 17-24.

Manalansan, M. F. (1996). Double minorities: Latino, Black, and Asian males who have sex with males. In R. C. Savin-Williams & K. M. Cohen (Eds.), The lives of lesbians, gays, and bisexuals (pp. 393-415). Fort Worth, TX: Harcourt Brace.

McCreary, D. R. (1994). The male role and avoiding femininity. Sex Roles, 31, 517-531.

McWhirter, D. P., & Mattison, A. M. (1984). The male couple: How relationships develop. Englewood Cliffs, NJ: Prentice-Hall.

Nardi, P. M. (1992). That's what friends are for: Friends as family in the gay and lesbian community. In K. Plummer (Ed.), Modern homosexualities (pp. 108-120). London: Routledge.

Nungesser, L. G. (1983). Homosexual acts, actors, and identities. New York: Praeger.

Pleck, J. H. (1981). The myth of masculinity. Cambridge, MA: MIT Press.

Rosario, M., Meyer-Bahlburg, H. F. L., Hunter, J., & Exner, T. M. (1997). The psychosocial development of urban lesbian, gay, and bisexual youths. The Journal of Sex Research, 33, 113-126.

Rotheram-Borus, M. J. (1994). Factors mediating changes in sexual HIV risk behavior among gay and bisexual male adolescents. American Journal of Public Health, 84, 1938-1946.

Rotheram-Borus, M. J., Reid, H., Rosario, M., & Kasen, S. (1995). Determinants of safer sex patterns among gay/bisexual male adolescents. Journal of Adolescence, 18, 3-15.

Savin-Williams, R. C. (1990). Gay and lesbian youth: Expressions of identity. Washington, DC: Hemisphere.

Savin-Williams, R. C. (1996). Memories of childhood and early adolescent sexual feelings among gay and bisexual boys: A narrative approach. In R. C. Savin-Williams & K. M. Cohen (Eds.), The lives of lesbians, gays, and bisexuals: Children to adults. Fort Worth, TX: Harcourt Brace & Co.

Savin-Williams, R. C. (1998). "... and then I became gay:" Young men's stories. New York: Routledge.

Sears, J. T. (1991). Growing up gay in the South: Race, gender, and journeys of the spirit. New York: Harrington Park Press.

Shidlo, A. (1994). Internalized homophobia. In B. Green & G. M. Herek (Eds.), Lesbian and gay psychology (pp. 176-205). Thousand Oaks, CA: Sage.

Stone, V. E., Catania, J. A., & Binson, D. (1999). Measuring change in sexual behavior concordance between survey measures. The Journal of Sex Research, 36, 102-108.

Tripp, C. A. (1975). The homosexual matrix. New York: McGraw-Hill.

Troiden, R. R. (1989). The formation of homosexual identities. Journal of Homosexuality, 17, 43-73.

Weinberg, T. S. (1978). On "doing" and "being" gay: Sexual behavior and homosexual male self-identity. Journal of Homosexuality, 4, 143-156.

Manuscript accepted December 7, 1999

Eric M. Dube Cornell University

Address correspondence to Eric M. Dube, 174 North Passaic Avenue, Chatham, NJ 07928; e-mail: [e-mail].