Partner-Specific Relationship Characteristics and Condom Use Among Young People With Sexually Transmitted Diseases.

Consistent condom use is a primary goal of current public health efforts to prevent sexually transmitted diseases (STDs). STD prevention efforts are especially targeted toward young people because of high infection rates among this population. Promotion of condom use as an habitual component of sexual activity has been at least partially successful: Nearly 60% of young people report condom use at last coitus (Warren et al., 1988). This remarkable level of condom use--three times greater than that reported in the early and mid 1980s--speaks for the substantial progress made in understanding the psychosocial correlates of condom use (Peipert et al., 1997; Pleck, Sonenstein, & Ku, 1993). However, risky sexual behaviors remain common and STD rates among young people remain high, suggesting that current approaches to condom promotion may have important limitations (Kelley & Kalichman, 1995).

One potential area for additional research is in understanding the influence of relationship characteristics on consistent condom use (Sheeran, Abraham, & Orbell, 1999). A number of studies show that various aspects of relationships are related to less frequent condom use (see Misovich, Fisher, & Fisher, 1997, for a comprehensive, recent review). Within diverse samples--including heterosexual adults, men who have sex with men, injection drug users, and heterosexual college students--condom use is less frequent in relationships characterized in some fashion that indicates relative relationship exclusivity and durability. Main, primary, and steady are terms commonly used in this context (Misovich et al., 1997; Sheeran et al., 1999). Condom use changes as the dynamics of a relationship change with time. Within a new partnership, condom use may be a tacit acknowledgment of lack of familiarity with the new partner. If the sexual relationship endures over a period of a few weeks (i.e., a new partner becomes an established partner), condom use may be reduced or stopped (Howard, Fortenberry, Blythe, Zimet, & Orr, 1999; Plichta, Weisman, Nathanson, Ensminger, & Robinson, 1992). Condom use is more likely, for example, with a new partner than with an established partner (Catania et al., 1989; Seidman, Mosher, & Aral, 1992; Wingood & DiClemente, 1998). Concurrent sexual relationships may also play an important role. Condom use is more likely with a side or secondary partner than with a main partner (Lansky, Thomas, & Earp, 1998). As noted in an extensive recent meta-analysis of psychosocial factors associated with condom use, the association of partner type and condom use is both consistent and strong (Sheeran et al., 1999).

Other studies use relationship duration as a marker for the relative stability of the sexual partnership. In general, these studies show that condom use and relationship duration are inversely related: Condom use becomes less consistent or is abandoned over a period of weeks or months (Ku, Sonenstein, & Pleck, 1994; Langer, Zimmerman, & Katz, 1994). Better known partners--including those to whom one is romantically attached--are assumed to be of low risk (Maticka-Tyndale, 1991, 1992; Misovich et al., 1997; Norris & Ford, 1999; Walter, Vaughn, Ragin, Cohall, & Kasen, 1994). This may account for the "sawtooth" pattern of condom use in sequential sexual relationships: high levels during the early weeks or months of a new relationship, decline in frequency of use, and resumption of high levels of use with a subsequent new partner (Ku et al., 1994). High levels of condom use are reinstituted in subsequent sexual relationships, but usage again declines over time. The period of time required for condom use to decrease is not well studied but may be as little as 21 days for some young people (Howard et al., 1999).

Coital frequency is another characteristic of sexual partnerships that influences condom use. A number of studies during the past 20 years show that condom use is less likely in relationships with higher coital frequency (Morrison, 1985). Coital frequency and condom use were correlated - 0.18 when averaged across multiple studies in a recent meta-analysis (Sheeran et al., 1999). High coital frequency may be a surrogate marker for relationships characterized by high levels of intimacy, sexual attraction, and abundant opportunity. Assumptions of sexual exclusivity may reduce perceived risk of STD. The couple may then choose contraceptive methods that are not physical barriers, do not imply lack of confidence in a partner's fidelity, and do not interfere with sexual pleasure (Gilmore, DeLamater, & Wagstaff, 1996; Hynie, Lydon, Cote, & Wiener, 1998).

Motivations for sexual intercourse may also influence condom use with a given partner (Hoffman & Bolton, 1997). Cooper, Shapiro, and Powers (1998) developed multi-item scales representing six key motivations for sex: Enhancement (of self), Intimacy, Coping, Self-Affirmation, Partner Approval, and Peer Approval. Intimacy motivations for sex were significantly associated with both more frequent coitus and less frequent condom use among a large, multi-ethnic sample of young people (Cooper et al., 1998). The magnitude of this relationship was markedly attenuated, however, by interactions with relationship status (i.e., in an exclusive sexual relationship for the previous 6 months).

The studies cited have several limitations. First, most rely on a limited number of measures of relationship characteristics--usually one or two. The complex associations of sexual pleasure, love, intimacy, fertility, and power suggest the need for a more detailed examination of relationship characteristics and condom use. Other research also typically depends on simple descriptors such as main or secondary to capture elements of commitment and relative relationship durability. The criteria by which researchers or subjects apply these descriptors to a specific relationship are seldom made explicit. Thus, the meaning of differences in condom use between main and secondary partners is usually left to the reader's interpretation. Direct assessment of the perceived quality of the relationship itself--separate from descriptive characteristics--has not been done in previous research. Finally, almost none of the existing research addresses the influence of partnership characteristics on a partner-specific basis.

We addressed potential influences of relationship characteristics on consistent condom use in a high-risk sample of young people with STDs. Each subject reported on a variety of characteristics of his or her relationship with each recent sexual partner. Measures of relationship quality and reasons for sex were developed in addition to more commonly used measures of relationship characteristics. The objective of this study is to assess associations of a variety of partner-specific relationship characteristics with consistent condom use among young people.

METHODS

Study Design

Subjects were 297 young people with gonorrhea, chlamydia, trichomonas, or nongonococcal urethritis (or a sexual contact of a person with one of these infections). These diseases were chosen to allow focus on curable STD among young people. All subjects were enrolled at the time of their STD treatment from a large STD clinic or at one of three neighborhood adolescent health clinics in a large city in the Midwest. Sociodemographic data for the study population are presented in Table 1.

Table 1. Sociodemographic and Behavioral Characteristics

Characteristic                      Value

Ethnicity (% African American)    75.1
Gender (% women)                  74.4
Age                               17.4 (1.9)(a)
Age of sexual debut               13.8 (1.9)(a)
Coital events, past 2 months(b)   12.0 (18.2)(a)
Number of sexual partners
  One                             199 (67.0)(c)
  Two                              62 (20.9)(c)
  Three                             23 (7.7)(c)
  Four                              13 (4.4)(c)

Note. N = 297.

(a) M (SD). (b) Summed across partners. (c) N(%).

Data Collection

At enrollment, all subjects completed a structured face-to-face interview with a research assistant. Items assessing behavioral frequencies were framed in terms of the previous 2 months. The interviewer identified each of the four most recent partners (within the past 2 months) by first name or initials. Subsequent items assessed partner-specific coital frequency and condom use, as well as a number of characteristics of the relationship.

Condom use was expressed as a proportion defined on the basis of two items: the total number of coital events and the number of coital events in which a condom was used. The distribution of this variable was unusual in that about one third of the partnerships involved no condom use, one third used condoms all the time, and one third were intermittent users (Table 2). Due to the distribution of the percentage of condom use, linear regression models did not provide an adequate fit for the data. Initially, we examined condom use as a three-category ordinal measure (Never, Intermittent, or Consistent). However, descriptive bivariate analyses of relationships with the independent variables showed that these relationships were not ordinal, and that partnerships in which condoms were used intermittently were similar to those in which condoms were never used. Therefore, all of the statistical analyses were performed after condom use percentage was categorized as Never/Intermittent versus Consistent (i.e., condom use at each coitus).

Table 2: Behavioral Relationship Characteristics

Characteristic                            Value

Relationship duration, in months          26.9 (49.0)(a)
Sexual relationship duration, in months    9.5 (13.8)(a)
Use of alcohol/drugs before sex           38.8(b)
Cohabitation                               9.5(b)
Child with partner                         6.6(b)
Condom use category (%)
   Consistent                             36.0
   Intermediate                           30.0
   Never                                  34.0

Note. N = 444.

(a) M (SD). (b) % yes.

Our approach to the measurement of condom use is an important aspect of this and all studies related to condom use. We chose a partner-specific consistency measure covering a 2-month interval. This time interval was chosen to optimize reliable recall of coitus and of condom use as well as to include a reasonable number of coital events (Catania, Gibson, Chitwood, & Coates, 1990). Some may argue that defining consistency as condom use at each coitus (i.e., 100% use) is an unnecessarily stringent behavioral standard. However, this definition is used in other research studies of adolescent condom use (DiClemente et al., 1996), and is consistent with public health recommendations for STD prevention.

Twelve independent variables were chosen to represent diverse aspects of sexual relationships. These variables were chosen to represent the complex interactions of interpersonal relationships, individual motives, and behavioral opportunity that may enhance or suppress a specific behavior such as condom use.

Partner-specific Relationship Quality was assessed by a 5-item scale developed for the current research. Items reflected the degree to which the subject reported a strong emotional relationship with the partner, enjoyed spending time with the partner, considered the partner to be a very important person, felt happy with the partner, and understood the partner as a person. Cronbach's alpha was 0.88 for women subjects and 0.94 for men. Higher scale scores reflect more positive relationship quality.

Partner-specific Reasons for Sex was assessed by three subscales. Intimacy Reasons for Sex was comprised by four items (to strengthen the relationship, to bring the couple closer together, to feel needed, and to feel loved). Cronbach's alpha was 0.78 and 0.77 for women and men, respectively. Sexual Desire Reasons for Sex was based on three items (partner horny, subject horny, and because sex feels good). Cronbach's alpha was 0.69 for women and 0.66 for men. Finally, Thrill Reasons for Sex was comprised by two items (just to get laid and just for thrills). Cronbach's alpha was 0.73 for women and 0.78 for men. Although differing in some respects, these subscales are similar to measures developed in other studies (Cooper et al., 1998; Hoffman & Bolton, 1997).

Two partner-specific measures of relationship durability were used: overall relationship duration assessed as the length of time (in months) the subject knew the partner, and sexual relationship duration assessed as the length of time (in months) since first coitus with a specific partner. Coital frequency was measured as the number of coital events in the previous two months. New partner reflected classification of the partner as new (no/yes). Cohabitation (no/yes) assessed whether the subject lived with the partner in the previous two months. Child with partner asked if the partner is father/mother of the subject's child (no/yes). Sexual coercion by the partner (no/yes) and use of alcohol/drugs before sex (no/yes) each reflect circumstantial aspects of sexual relationships that might interfere with consistent condom use.

Subject-specific characteristics that were examined were age (in years), ethnicity (non-African-American versus African American), gender, age of sexual debut (in years), and number of sexual partners in the previous two months. These individual-level characteristics are linked to variation in condom use in a number of earlier studies (Ku et al., 1994; Sheeran et al., 1999).

Each subject contributed between one and four observations (444 total partnerships) to the study. The distribution of partnerships by subject can be found in Table 1. Sixty-seven percent of the subjects reported one sexual partner in the previous 2 months.

Statistical Analysis

Bivariate analyses were conducted using simple logistic regression and generalized estimating equations (GEE) to adjust for within-subject correlations. Subject-specific and partner-specific variables that showed at least a moderate bivariate association with consistent condom use (p [is less than] 0.20) were included in stepwise and backward logistic regression procedures to determine sets of independent predictors for consistent condom use, relative to nonuse/intermittent use. Variables contained in either model were then combined to determine if a single better model could be identified. Interactions among the explanatory variables were also examined. All multivariable models were fit using generalized linear models with a logit link function and a binomial error term (McCullagh & Nelder, 1989). Standard errors and p-values were adjusted for within-subject correlation using GEE (Zeger & Liang, 1986; Zeger, Liang, & Albert, 1988). Statistical significance was defined as p [is less than] 0.05.

RESULTS

Subject-specific bivariate associations with consistent condom use are presented in Table 3. Men and African Americans were about twice as likely to report consistent condom use in a relationship. The odds of consistent condom use decreased with age and increased with the number of recent partners. Age of sexual debut was not associated with condom use.

Table 3. Subject-Specific Bivariate Associations with Consistent Condom Use

Variable                               Odds ratio (95% CI)    p

Age(a)                                 0.853 (0.759, 0.957)   0.007
Age at sexual debut(a)                 0.962 (0.860, 1.077)   0.50
Number of partners (past 2 months)(b)  1.117 (1.001, 1.247)   0.048
Male gender                            1.972 (1.270, 3.062)   0.003
African American ethnicity             2.192 (1.256, 3.826)   0.006

Note. Bivariate logistic regression; adjusted for within subject correlations by generalized estimating equations.

(a) Odds ratio is presented for each additional year of age. (b) Odds ratio is presented for each additional partner.

Partner-specific bivariate associations with consistent condom use are presented in Table 4. Consistent condom use was significantly more likely with partners characterized as new and was significantly less likely with cohabitating partners and when the couple had a child. The odds of consistent condom use decreased as the number of coital events within the partnership increased.

Table 4. Partner-Specific Bivariate Associations with Consistent Condom Use

Variable                          Odds ratio (95% CI)    p

Relationship quality              0.834 (0.744, 0.935)     0.002
Reasons for sex scales(a)
  Emotional                       0.879 (0.838, 0.923)   < 0.001
  Sexual desire                   0.946 (0.862, 1.038)     0.24
  Thrill                          1.117 (0.978, 1.276)     0.10
Relationship duration(b)          0.999 (0.995, 1.003)     0.55
Sexual relationship duration(b)   0.981 (0.958, 1.004)     0.098
Number of coital events           0.853 (0.788, 0.924)   < 0.001
  (previous 2 months)(c)
New partner                       2.040 (1.374, 3.028)   < 0.001
Cohabitation                      0.267 (0.108, 0.659)     0.004
Child with partner                0.266 (0.091, 0.781)     0.016
Sexual coercion                   0.494 (0.210, 1.162)     0.11
Drug/alcohol use                  0.775 (0.499, 1.205)     0.26

Note. Bivariate logistic regression; adjusted for within subject correlations by generalized estimating equations.

(a) Odds ratio is presented for each unit on the scale. (b) Odds ratio is presented for each additional month. (c) Odds ratio is presented for each additional event.

In terms of relationship quality and reasons for sex, consistent condom use was less likely in relationships with higher Relationship Quality and higher Intimacy Reasons for Sex. There were no statistically significant associations between condom use and relationship duration, relationships involving sexual coercion or drug/alcohol use, or the two Reasons for Sex subscales reflecting sexual desire and thrill.

The multivariable modeling resulted in two explanatory models for these data. The initial model derived from the backwards selection procedure included three factors: male gender, number of coital events, and having a child in the partnership. The model obtained using the stepwise selection method contained four variables: male gender, age, number of coital events, and Relationship Quality. Alternative models were examined that included male gender, number of coital events, and all subsets of the three remaining variables that were included in either of the original models. The two original models were determined to yield the best fit among these candidate models. Next, because adolescent men and women often show gender-specific behaviors, interactions with gender were added to each model. Only the gender by Relationship Quality interaction was found to be significant.

The factors exhibiting independent associations with consistent condom use in the first multiple logistic regression model are shown in Table 5. This is the more parsimonious of the two models but it provides only a partial picture of condom use in the study population, particularly since one of the three predictors, having a child in the partnership, applies to a small proportion of couples. At the subject level, consistent condom use was associated with male gender. Men were more than twice as likely to report consistent condom use. At the partner level, consistent condom use was over 3 times more likely when there was no child with the partner. A higher number of coital events was associated with reduced odds of consistent condom use.

Table 5. Predictors of Consistent Condom Use--Multiple Logistic Regression Model 1

                                        95% Confidence
Variable                   Odds ratio      interval         p
Number of coital events      0.833      0.768, 0.903     < 0.001
  (previous 2 months)(a)
Male gender                  2.316      1.393, 3.852       0.001
Child with partner           0.271      0.082, 0.898       0.033

Note. Adjusted for within-subject correlation by generalized estimating equations.

(a) Odds ratio is presented for each additional event. (b) Odds ratio is presented for each additional year of age. (c) Odds ratio is presented for each unit on the scale.

The factors exhibiting independent associations with consistent condom use in the second multiple logistic regression model are shown in Table 6. The model includes two subject-specific factors, age and male gender, and two partner-specific factors, number of coital events and Relationship Quality. In addition, a significant gender by Relationship Quality interaction was included. Each additional year of age resulted in a 14.5% decrease in the odds of consistent condom use. Similarly, each additional coital event within a partnership resulted in an 11.5% decrease in the odds of consistent condom use. The effect of Relationship Quality differed by gender. For men, the odds of consistent condom use decreased by 14.4% with each unit increase in the Relationship Quality scale. In contrast, for women the odds ratio was approximately one (OR = 0.98), indicating that relationship quality had little effect on consistent condom use.

Table 6. Predictors of Consistent Condom Use--Multiple Logistic Regression Model 2

                                               95% Confidence
Variable                         Odds ratio       interval

Number of coital events            0.885        0.825, 0.949
  (previous 2 months)(a)
Age(b)                             0.845        0.745, 0.959
Male gender                     See Figure 1
Relationship quality(c)
  For men                          0.856        0.783, 0.937
  For women                        0.983        0.921, 1.049
Gender by quality interaction

Variable                              p

Number of coital events           < 0.001
  (previous 2 months)(a)
Age(b)                              0.009
Male gender                         0.001
Relationship quality(c)           < 0.001
  For men
  For women
Gender by quality interaction       0.012

Note. Adjusted for within-subject correlation by generalized estimating equations.

(a) Odds ratio is presented for each additional event. (b) Odds ratio is presented for each additional year of age. (c) Odds ratio is presented for each unit on the scale.

The estimated odds ratios for men to women across the range of the Relationship Quality scale are presented in Figure 1. For the lowest scale score, men are 10 times more likely to report consistent condom use than are women. At the upper end of the scale the odds ratio is approximately one, indicating little difference in consistent condom use between genders for higher quality relationships.

[Figure 1 ILLUSTRATION OMITTED]

DISCUSSION

The data collected for the present study attempt to explain consistent condom use within the context of sexual couples' interpersonal relationships. This level of detail allowed us to examine factors related to specific relationships while controlling for individual characteristics associated with condom use in previous research.

Several characteristics of the interpersonal relationship independently affected consistent condom use. Taken together, these characteristics suggest that the complex interpersonal context of sex influences the condom use within a given relationship. Factors that imply some degree of enduring connection--having a child together or perceiving the relationship in emotionally positive terms--significantly reduce the odds of consistent condom use. In addition, these factors may be perceived differently by men and women as illustrated by the gender differences in the association of Relationship Quality and consistent condom use. In relationships characterized by fewer emotional and affiliative qualities, men are 10 times more likely than women in similar relationships to consistently use a condom. In relationships marked by more emotional connections, consistent condom use is the same for men and women. This finding is consonant with the findings of other research that men choose partners based on factors such as sexual drive and reputation, while women use relational ideals (Hynie et al., 1998; Kirkman, Rosenthal, & Smith, 1998; Maticka-Tyndale, 1992).

Coital frequency was an additional relationship characteristic related to consistent condom use. This is a finding demonstrated in several other studies (Sheeran et al., 1999). Increased coital frequency could simply reflect increased opportunity for condom nonuse for at least one coital event. The often-cited disadvantages of condom use--inconvenience, messiness, and interference with sexual pleasure--might be magnified when coitus is more frequent with a specific partner (Jadack, Fresia, Rompalo, & Zenilman, 1997). Perceived STD risk of a partner may also decrease as coital frequency increases. Young people perceive greater STD risk and greater intention to use condoms with casual than with main sex partners (Ellen et al., 1998). In our analyses, identification of a partner as "new" was associated with consistent condom use in bivariate but not in multivariate analyses. Coital frequency itself may be a marker for some qualitative aspect of the interpersonal relationship. It is also possible that higher coital frequency is associated with greater perceived need for contraception rather than STD prevention. This would lead to the use of nonbarrier contraceptives and decreased emphasis on condom use within the relationship (Hammer, Fisher, Fitzgerald, & Fisher, 1996).

These data are consistent with those of previous studies showing that condom use is a function of certain characteristics of the sexual partnership. Condom use may be particularly difficult to negotiate in relationships characterized by economic dependence or fear of physical violence (Wingood & DiClemente, 1998; Worth, 1989). However, we did not find that sexual coercion was associated with decreases in consistent condom use. In agreement with other research, we found that condom use is more likely with new partners or with partners characterized as side or secondary (Catania et al., 1989; Joffe et al., 1992; Lansky et al., 1998; Marin, Gomez, & Hearst, 1993; Plichta et al., 1992; Seidman et al., 1992). Women in exclusive sexual relationships perceive themselves to be at low risk for STDs, have low intentions to use condoms, and are less likely to use condoms (St. Lawrence et al., 1998). One study of young people found that cohabitation and emotional closeness to a partner were each associated with a reduced likelihood of developing intentions to use condoms (Santelli et al., 1996).

Several limitations should be considered in evaluating these data. First, the sample was comprised by a relatively homogenous racial group within a single urban area. Caution is needed in generalizing to other populations.

Second, condom use was measured as the partner-specific proportion of condom-protected events. This measure was then dichotomized to less than 100% use versus 100% use. Misclassification bias could occur if subjects underreported the total number of coital events or overreported the number of coital events that were condom protected, or both. Underreporting of coital events is likely, especially for higher levels of coital frequency (Fortenberry, Cecil, Zimet, & Orr, 1997). However, coital frequencies reported in the current study were fairly low and are likely to be relatively accurate. We can find no studies to suggest that reports of coital frequency or levels of condom use are biased by characteristics of the subjects' sexual relationships.

Our data suggest that the level of consistent condom use within a given adolescent population is dependent on the characteristics of the sexual interpersonal relationships within that population. While the marked increase in condom use among young people during the past two decades represents a remarkable public health success, relatively high STD rates among young people support the need for additional efforts to improve prevention. These efforts almost certainly need to give greater attention to the complex interpersonal dynamics of sexual relationships if additional progress is expected.

REFERENCES

Catania, J. A., Dolcini, M. M., Coates, T. J., Kegeles, S. M., Greenblatt, R. M., Puckett, S., Corman, M., & Miller, J. (1989). Predictors of condom use and multiple partnered sex among sexually-active adolescent women: Implications for AIDS-related health interventions. The Journal of Sex Research, 26, 514-524.

Catania, J. A., Gibson, D. R., Chitwood, D. D., & Coates, T. J. (1990). Methodological problems in AIDS behavioral research: Influences on measurement error and participation bias in studies of sexual behavior. Psychological Bulletin, 108, 339-362.

Cooper, M. L., Shapiro, C. M., & Powers, A. M. (1998). Motivations for sex and risky sexual behavior among young people and young adults: A functional perspective. Journal of Personality and Social Psychology, 75, 1528-1558.

DiClemente, R. J., Lodico, M., Grinstead, O. A., Harper, G., Rickman, R. L., Evans, P. E., & Coates, T. J. (1996). African-American young people residing in high-risk urban environments do use condoms: Correlates and predictors of condom use among young people in public housing developments. Pediatrics, 98, 269-278.

Ellen, J. M., Adler, N. E., Dunlop, M. B. V., Gurvey, J. E., Millstein, S. G., & Tschann, J. M. (1998). Perceived risk for STDs and partner-specific condom intentions [Abstract]. Pediatric Research, 43, 4A.

Fortenberry, J. D., Cecil, H., Zimet, G. D., & Orr, D. P. (1997). Concordance between self-report questionnaires and coital diaries for sexual behaviors of adolescent women with sexually transmitted diseases. In J. Bancroft (Ed.), Researching sexual behavior (pp. 237-249). Bloomington, IN: Indiana University Press.

Gilmore, S., DeLamater, J., & Wagstaff, D. (1996). Sexual decision making by inner city black adolescent males: A focus group study. The Journal of Sex Research, 33, 363-371.

Hammer, J. C., Fisher, J. D., Fitzgerald, J. F., & Fisher, W. A. (1996). When two heads aren't better than one: AIDS risk behavior in college-age couples. Journal of Applied Social Psychology, 26, 375-397.

Hoffman, V., & Bolton, R. (1997). Reasons for having sex and sexual risk-taking: A study of heterosexual STD clinic patients. AIDS Care, 9, 285-296.

Howard, M. M., Fortenberry, J. D., Blythe, M. J., Zimet, G. D., & Orr, D. P. (1999). Patterns of sexual partnerships among adolescent females. Journal of Adolescent Health, 24, 300-303.

Hynie, M., Lydon, J. E., Cote, S., & Wiener, S. (1998). Relational sexual scripts and women's condom use: The importance of internalized norms. The Journal of Sex Research, 35, 370-380.

Jadack, R. A., Fresia, A. F., Rompalo, A. M., & Zenilman, J. (1997). Reasons for not using condoms of clients at urban sexually transmitted diseases clinics. Sexually Transmitted Diseases, 24, 402-408.

Joffe, G. P., Foxman, B., Schmidt, A. J., Farris, K. B., Carter, R. J., Neumann, S., Tolo, K. A., & Walters, A. M. (1992). Multiple partners and partner choice as risk factors for sexually transmitted disease among female college students. Sexually Transmitted Diseases, 19, 272-278.

Kelley, J. A., & Kalichman, S. C. (1995). Increased attention to human sexuality can improve HIV-AIDS prevention efforts: Key research issues and directions. Journal of Consulting & Clinical Psychology, 63, 907-918.

Kirkman, M., Rosenthal, D., & Smilth, A. M. A. (1998). Adolescent sex and the romantic narrative: Why some young heterosexuals use condoms to prevent pregnancy but not disease. Psychology, Health and Medicine, 3, 355-370.

Ku, L., Sonenstein, F. L., & Pleck, J. H. (1994). The dynamics of young men's condom use during and across relationships. Family Planning Perspectives, 26, 246-251.

Langer, L. M., Zimmerman, R. S., & Katz, J. A. (1994). Which is more important to high school students: Preventing pregnancy or preventing AIDS? Family Planning Perspectives, 26, 154-159.

Lansky, A., Thomas, J. C., & Earp, J. A. (1998). Partner-specific sexual behaviors among persons with both main and other partners. Family Planning Perspectives, 30, 93-96.

Marin, B. V. O., Gomez, C. A., & Hearst, N. (1993). Multiple heterosexual partners and condom use among Hispanics and non-Hispanic Whites. Family Planning Perspectives, 25, 170-174.

Maticka-Tyndale, E. (1991). Sexual scripts and AIDS prevention: Variations in adherence to safer-sex guidelines by heterosexual young people. The Journal of Sex Research, 28, 45-66.

Maticka-Tyndale, E. (1992). Social construction of HIV transmission and prevention among heterosexual young adults. Social Problems, 39, 238-252.

McCullagh, P., & Nelder, J. A. (1989). Generalized linear models (2nd ed.). London: Chapman and Hall.

Misovich, S. J., Fisher, J. D., & Fisher, W. A. (1997). Close relationships and elevated HIV risk behavior: Evidence and possible underlying psychological processes. Review of General Psychology, 1, 72-107.

Morrison, D. M. (1985). Adolescent contraceptive behavior: A review. Psychological Bulletin, 98, 538-568.

Norris, A. E., & Ford, K. (1999). Sexual experiences and condom use of heterosexual, low-income African-American and Hispanic youth practicing relative monogamy, serial monogamy, and nonmonogamy. Sexually Transmitted Diseases, 26, 17-25.

Peipert, J. F., Domagalski, L., Boardman, L., Daamen, M., McCormack, W. M., & Zinner, S. H. (1997). Sexual behavior and contraceptive use. Changes from 1975 to 1995 in college women. Journal of Reproductive Medicine, 42, 651-657.

Pleck, J. H., Sonenstein, F. L., & Ku, L. (1993). Changes in adolescent males' use of and attitudes toward condoms, 1988-1991. Family Planning Perspectives, 25, 106-109.

Plichta, S. B., Weisman, C. S., Nathanson, C. A., Ensminger, M. E., & Robinson, J. C. (1992). Partner-specific condom use among adolescent women clients of a family planning clinic. Journal of Adolescent Health, 13, 506-511.

Santelli, J. S., Kouzis, A. C., Hoover, D. R., Polacsek, M., Burwell, L. G., & Celentano, D. D. (1996). Stage of behavior change for condom use: The influence of partner type, relationship and pregnancy factors. Family Planning Perspectives, 28, 101-107.

Seidman, S. N., Mosher, W. D., & Aral, S. O. (1992). Women with multiple sexual partners: United States, 1988. American Journal of Public Health, 82, 1388-1394.

Sheeran, P., & Abraham, C. (1994). Measurement of condom use in 72 studies of HIV-preventive behaviour: A critical review. Patient Education and Counseling, 24, 199-216.

Sheeran, P., Abraham, C., & Orbell, S. (1999). Psychosocial correlates of heterosexual condom use: A meta-analysis. Psychological Bulletin, 125, 90-132.

St. Lawrence, J. S., Eldridge, G. D., Reitman, D., Little, C. E., Shelby, M. C., & Brasfield, T. L. (1998). Factors influencing condom use among African American women: Implications for risk reduction interventions. American Journal of Community Psychology, 26, 7-28.

Walter, H. J., Vaughn, R. D., Jr., Ragin, D. R., Cohall, A. T., & Kasen, S. (1994). Prevalence and correlates of AIDS-related behavioral intentions among urban minority high school students. AIDS Education & Prevention, 4, 339-350.

Warren, C. W., Santelli, J. S., Everett, S. A., Kann, L., Collins, J. L., Cassell, C., Morris, L., & Kolbe, L. J. (1998). Sexual behavior among U.S. high school students, 1990-1995. Family Planning Perspectives, 30, 170-172.

Wingood, G. M., & DiClemente, R. J. (1998). Partner influences and gender-related factors associated with noncondom use among young adult African American women. American Journal of Community Psychology, 26, 29-51.

Worth, D. (1989). Sexual decision-making and AIDS: Why condom promotion among vulnerable women is likely to fail. Studies in Family Planning, 20, 297-307.

Zeger, S., Liang, K., & Albert, P. (1988). Models for longitudinal data: A generalized estimating equation approach. Biometrics, 44, 1049-1060.

Zeger, S. L., & Liang, K-. Y. (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121-130.

Manuscript accepted October 25, 1999

Barry P. Katz, J. Dennis Fortenberry, Gregory D. Zimet, Margaret J. Blythe, and Donald P. Orr Indiana University School of Medicine

Funded by a grant from the National Institute of Allergy and Infectious Diseases (U19 AI31494). Presented in part at the meeting of the International Society for Sexually Transmitted Diseases Research, October 1997, Seville, Spain. The authors thank Patricia Brooks and Cathy Roberts for invaluable assistance with data collection and Tim Breen and Kathleen Bergan for their data management support.

Address correspondence to Barry P. Katz, Ph.D., Division of Biostatistics, Indiana University, 1050 Wishard Boulevard, RG 4101, Indianapolis, IN 46202-2872; e-mail: [e-mail].