Family Planning Perspectives, 1998, 30(6):276-280 & 303

Unprotected sexual intercourse among American teenagers has many serious public health and societal consequences, particularly the transmission of HIV and other sexually transmitted diseases (STDs). Among the 40,000-80,000 Americans who become infected with HIV each year, one in four are teenagers.,1In addition, adolescents account for one-third of all chlamydia and gonorrhea cases and one-10th of all syphilis cases in the United States.2

The risk of STD transmission among youths may be reduced by encouraging safer sexual behaviors, such as the use of condoms. Next to abstinence, consistent and correct condom use is the most effective way to prevent the transmission of HIV.3 Indeed, as part of the U.S. government's Healthy People 2000 initiative, the surgeon general's office has set a goal of increasing teenagers' use of condoms at last intercourse to 75%.4

Efforts to increase use of condoms among teenagers appear to have been at least moderately successful. Data from the National Survey of Adolescent Males (NSAM) indicate that between 1988 and 1995, the rate of condom use at last intercourse among 15-19-year-old males increased from 56% to 69%,5six percentage points shy of the surgeon general's goal. A similar increase in condom use occurred among adolescent females during the same time period, according to data from the 1988 and 1995 cycles of the National Survey of Family Growth.6

However, the majority of adolescent males do not use condoms consistently. Although the percentage of teenage males using condoms at every act of intercourse increased between 1988 and 1995, only 45% of teenage males in the 1995 NSAM sample reported using condoms consistently during the 12 months preceding the survey.7

Research indicates that attitudes and beliefs concerning the effectiveness and social consequences of condom use are major determinants of use.8 One study using 1988 NSAM data found that five factors were associated with consistency of condom use: the belief that males have some responsibility for preventing pregnancy; the frequency of worry about AIDS; anticipation that a partner would appreciate the respondent's willingness to use condoms; the perception that condoms are embarrassing; and the belief that condoms significantly reduce men's sexual pleasure.9

Analysis of data from the 1988 and 1991 waves of the NSAM determined that males aged 17.5-19 in 1991 were more likely than those in 1988 to endorse positive condom-related attitudes, and that these attitudes were significant predictors of young men's use of condoms.10 The study also found that changes in perceived reduction of sexual pleasure and in the female partner's appreciation of a male's condom use were associated with changes in condom use between the two samples, while multivariate analyses showed that decreased worry about AIDS was associated with decreased condom use.11

Using data from the NSAM, we explore the role of attitudes and beliefs in explaining the observed increase in condom use between 1988 and 1995 among males aged 15-19. Given the trends in condom use, we hypothesize that young males in general in the 1995 sample will have adopted stronger attitudes concerning male contraceptive responsibility and will have more favorable attitudes about condoms than those in the earlier sample.

Methods

Data

We analyzed data from two independent waves of the NSAM, administered in 1988 and 1995. The NSAM is a nationally representative survey of never-married, noninstitutionalized 15-19-year-old men residing in the contiguous United States. Each cohort was drawn as a multistage area-probability sample that overrepresented blacks and Hispanics. The 1988 survey had a sample size of 1,880, with a response rate of 74% among eligible respondents, while the 1995 survey had a sample size of 1,729 and a response rate of 75%.12

Collected data included partner-by-partner sexual and contraceptive histories, demographic and family background variables, and HIV- and pregnancy-related attitudes and knowledge. Postsampling weights were developed in 1988 to match the March 1987 Current Population Survey13 and in 1995 to match the preliminary census projections of the civilian noninstitutionalized population.14

Because the NSAM was designed specifically for the purpose of studying the sexual and contraceptive behavior of adolescent males, it contains a wealth of information about both sexual and contraceptive behaviors and attitudes. However, since all information was collected from males, very little could be ascertained about the attitudes and behaviors of their girlfriends and sexual partners.

The analytic sample in the current study was limited to adolescent males who engaged in sexual intercourse with a female at least once in the 12 months preceding the survey. In 1988, 60% of males were sexually active, compared with 55% in 1995.15This decrease in sexual activity is analyzed in-depth elsewhere.16

To identify relationships between condom-related attitudes and behaviors and differences by cohort and race, we merged data from the 1988 and 1995 waves into a pooled sample. This sample included all sexually active males aged 15-19 in either of the two waves. The combined sample was composed of 2,295 males, 1,263 of whom were interviewed in 1988 and 1,032 of whom were interviewed in 1995. Postsampling weights were applied to all analyses.

Measures

Condom use. Consistency of condom use, defined as the percentage of times a respondent used a condom, either alone or in combination with other contraceptive methods, during heterosexual intercourse over the 12 months preceding the survey, was the dependent variable in all analyses. This variable was measured by compiling partner-by-partner data to obtain an estimate of total condom usage over the last year: Respondents reported the percentage of times they had used a condom with each of their last six partners during the previous year. Weighting each partner by the number of times the couple engaged in sexual intercourse resulted in a measure of condom-use consistency for each respondent.17 Condom-use consistency was coded with values between zero and one, inclusive (where, for example, 0.40=40%). For partners with whom the respondent reported having intercourse only once, consistency of condom use was coded as either zero or one (0% or 100%).

Condom-related attitudes. Following the approach used in previous analyses,18 we divided condom-related attitudes into three distinct categories: pregnancy prevention, AIDS avoidance and personal and social consequences.

Pregnancy prevention attitudes included three variables: the perceived likelihood that a female partner would get pregnant if the respondent used a condom (perceived condom ineffectiveness); the degree to which the respondent believes that if he made someone pregnant, he would feel like a "real man"; and a five-item index assessing the degree to which the respondent believes in male contraceptive responsibility.* The perceived condom ineffectiveness variable is scored on a scale ranging from one (no chance of pregnancy) to five (almost certain chance). The index of male contraceptive responsibility represents the average response to a series of five questions, each measuring a different component of male contraceptive responsibility. Items in this index are scored on a scale ranging from one to four, with higher values denoting more responsible attitudes. The "real man" variable is scored on a scale ranging from one (not at all) to four (a lot).

Three aspects of AIDS avoidance are measured in the NSAM: the frequency with which the respondent worries about AIDS; the respondent's perceived likelihood of HIV infection within the next five years; and the respondent's degree of AIDS denial. The first two variables are scored on a scale ranging from one to five, with higher values denoting greater concern about AIDS. The AIDS denial variable is an index composed of two questions that measure the respondent's views on AIDS prevalence and his perception of the benefit of using a condom to avoid HIV infection. The AIDS denial variable is scored on a scale from one to four, with higher values denoting increased denial.

Attitudes concerning personal and social consequences of condom use include the degree to which the respondent believes his partner would appreciate his use of a condom; the degree to which the respondent believes he would feel less pleasure if he used a condom; and a four-item index of condom-use embarrassment. (The index assesses embarrassment associated with buying, discussing and using condoms.) All three variables are scored on a scale ranging from one to five, with higher values corresponding to greater appreciation, less pleasure and increased embarrassment, respectively.

Results

Condom-Use Consistency

Overall, according to Table 1, condom-use consistency increased by 24% between 1988 and 1995. Sexually active males in 1988 reported using condoms in 56% of their sexual encountersin the 12 months prior to the survey, compared with a mean use-consistency rate of 69% in 1995.

The increase in condom-use consistency differed across racial and ethnic groups. The greatest increase occurred among white males. In this group, condom-use consistency increased from 54% in 1988 to 70% in 1995, an increase of 31%. In contrast, condom-use consistency among blacks increased by 17%, from 63% in 1988 to 73% in 1995. While whites had lower levels of condom-use consistency than did blacks in both 1988 and 1995, the difference in the rate of consistency between the two groups decreased over time. The smallest change in condom-use consistency, an increase of 12%, occurred among Hispanics.

Condom-Related Attitudes and Beliefs

Table 2 depicts the distribution of responses to the 17 condom-related attitude questions and the mean response for each question for both the 1988 and the 1995 cohorts. Chi-square tests were used to test for significant changes in the distribution of responses across cohorts.

Pregnancy prevention. Adolescent males in 1995 reported a significantly weaker belief in the role of condoms in preventing pregnancy than did those in 1988: The proportion of males who thought there was little or no chance that a new female partner would get pregnant if they always used a condom decreased from 71% in 1988 to 64% in 1995. Overall, adolescent males in 1995 were significantly less likely than their 1988 counterparts to hold attitudes suggesting contraceptive responsibility (mean index scores of 3.45 and 3.52, respectively; p<.01, not shown). Males in 1995 were significantly less likely than males in 1988 to "agree a lot" that they should ask whether their partner is practicing contraception, that they should help pay for their partner's pill prescription and that they should bring up the issue of contraception.

While the distribution of responses to the question on the extent to which causing a pregnancy would make the respondent feel like a "real man" changed significantly over time, this was the result of fewer males' choosing the more moderate responses of "a little" and "somewhat" and more males' choosing the extreme responses of "not at all" and "a lot"; the mean response score did not change between 1988 and 1995.

AIDS avoidance. The mean scores for AIDS worry and perceived AIDS risk among males surveyed in 1995 were significantly lower than the mean scores among males in 1988. However, the distributions of these scores were not significantly different between the two cohorts. Respondents in the 1995 sample were more likely than those in the earlier wave to disagree that AIDS is so uncommon that it is not a big worry, but there was no significant change in the overall degree of AIDS denial, and mean scores on the index of AIDS denial were not significantly different between males in 1995 and 1988 (1.34 vs. 1.31, not shown). However, the distribution of responses to the two items in this index each changed significantly over time, as respondents tended in 1995 to both agree or disagree less strongly with the item "using condoms to prevent the spread of AIDS is more trouble than it's worth" and to agree more that the disease is uncommon and therefore less worrisome than did respondents in 1988.

Personal and social consequences. Respondents in the 1995 sample were significantly more likely to believe that their partner would appreciate their use of a condom than were respondents in the 1988 sample: The proportion of males believing that there was a greater than 50% chance that their female partner would appreciate their use of a condom increased from 68% in 1988 to 78% in 1995.

Condom-use embarrassment was significantly lower among members of the 1995 cohort than among those in 1988 (mean index score of 2.22 in 1995, compared with 2.41 in 1988, p<.01, not shown). More than 40% of males in the 1995 sample indicated that there was "no chance" that they would feel embarrassed either buying a condom in a drugstore, discussing condoms with their partner or putting a condom on in front of their partner. Among respondents in 1988, the percent reporting "no chance" of embarrassment to items on this index ranged from 20% to 32%.

Respondents in 1995 were less likely to believe that decreased pleasure would accompany condom use than were 1988 respondents. Nearly 50% of 1995 respondents believed there was little or no chance that they would feel less pleasure using a condom, compared with 45% in 1988.

Multivariate Models

Predictors of condom-use consistency. Table 3 presents the results of the multivariate models estimating the relationship between condom-related attitudes and condom-use consistency. Because we measured attitudes subsequent to the time period during which condom use was measured, we cannot ascertain the extent to which condom-related attitudes influenced condom use or the extent to which condom use affected condom-related attitudes. While it is likely that attitudes and condom use influence each other, these models should be interpreted only as depicting the correlation between attitudes at the time of the survey and behavior during the 12 months preceding the survey.

Following the methodology employed in previous studies,19 we used a two-limit tobit model to analyze the relationship between condom attitudes and condom use. The two-limit tobit is appropriate in this case because of the clustering of the dependent variable at 0% and 100%. The table presents both the tobit coefficients and the marginal effects, which indicate the percentage-point change in condom-use consistency attributable to a one-unit change in the independent variable, evaluated at the means of the other independent variables.

The regression model in the first panel of Table 3 includes all nine measures of condom-related attitudes: six individual items and the three constructed indices. In addition, several independent controls were included in the model to better isolate the effects of these attitudes: the respondent's age and race; the level of education of his more-educated parent;§ whether he receives welfare benefits; whether the respondent lived in a female-headed household at age 14; his age at first intercourse; and the year in which the respondent was interviewed.

Two pregnancy prevention attitudes were significantly associated with consistent condom use. Males who held a strong belief in male contraceptive responsibility reported significantly higher levels of condom use in the 12 months prior to the interview than did those believing that men do not need to take much responsibility. In addition, there was a small but significant positive correlation between a respondent's feeling like a "real man" by causing a pregnancy and his condom-use consistency, indicating that higher rates of condom use were associated with the belief that causing a pregnancy enhanced masculine feelings. There was no significant association between perceived condom ineffectiveness and consistent condom use in this model.

Two AIDS avoidance attitudes were significantly associated with consistent condom use. The frequency of AIDS worry was positively associated with consistent condom use, while the degree of AIDS denial was negatively correlated with condom-use consistency. Perceived risk of contracting HIV was marginally associated with consistent condom use (p<.10).

Attitudes about the personal and social consequences of condom use also were significantly correlated with use consistency. The belief that a sex partner would appreciate condom use was positively associated with consistency, and embarrassment regarding discussing, purchasing or using a condom was negatively associated with consistency of condom use. In addition, the belief that condom use decreases pleasure was associated with less consistent condom use.

Being black, being younger, having a more educated parent, having been older at first intercourse and participating in the 1995 survey all were positively and significantly associated with consistent condom use.

Differences over time. Not all condom-related attitudes shown in Table 2 changed in a direction that corresponds with the increases in condom use depicted in Table 1. One possible explanation is that there was a change in the relationship between condom-related attitudes and condom use between 1988 and 1995.

To test for significant differences between survey years in the relationship between condom-use consistency and condom attitudes, we reestimated the original model and included a series of interaction terms between each attitudinal variable and a dummy variable indicating whether the respondent was part of the 1995 cohort. If there has been a significant change over time in the relationship between a condom-related attitude and condom-use consistency, the coefficient on the interaction term for the attitudinal variable should be significant.

In Table 3, the multivariate model including the significant interaction terms is presented in the second panel. The effect of each attitude on condom-use consistency in 1988 is indicated by the coefficient for that variable alone, while the effect in 1995 is determined by summing the coefficient for that attitude and the coefficient of its interaction term.** The sign of the coefficient for the interaction terms indicates the direction of change of the relationship between 1988 and 1995.

The interaction terms for condom ineffectiveness, male contraceptive responsibility and all three AIDS avoidance attitudes were not significantly different from zero and were dropped from the final model.

Among variables associated with pregnancy prevention, the interaction term for "feel like a 'real man'" was negative and marginally significant (p<.10), indicating that this attitude was somewhat more important in predicting condom use in 1988 than it was in 1995. The change in this relationship suggests that levels of condom-use consistency should be lower in 1995 than in 1988, all other variables being constant.

The interaction term for partner appreciation by survey year also was negative and marginally significant (p<.10), indicating that while increased partner appreciation was associated with more consistent condom use in both 1988 and 1995, the relationship was somewhat weaker in 1995. Therefore, although males in the 1995 cohort perceived greater partner appreciation for condom use than did males in the 1988 cohort, this attitude does not explain the observed increase in condom-use consistency between the two cohorts.

The interaction term for condom-use embarrassment by survey year was significant and positive, indicating that while embarrassment was negatively associated with condom use in 1988, there was no significant relationship between these variables in 1995. Given the lower levels of condom-use embarrassment assessed among the 1995 respondents than in the earlier group, this finding predicts higher levels of condom use among 1995 respondents than among those in 1988, all other variables being constant.

Finally, the significant negative coefficient for the interaction between feeling less pleasure when using a condom and survey year indicates that the relationship between this variable and consistent condom use was much stronger in 1988 than it was in 1995. Thus, while pleasure reduction attitudes changed over time in a direction that predicts higher levels of condom use, the change in the relationship between this attitude and condom use actually predicts lower levels of condom-use consistency in 1995 compared with 1988, all other variables being constant.

Discussion

Three conclusions emerge from this analysis. First, consistent condom use among young, sexually active males increased significantly between 1988 and 1995. Coupled with a 9% decrease in the percentage of adolescent males who were sexually active between 1988 and 1995,20this finding indicates that young men are increasingly taking precautions to avoid unwanted pregnancy and the transmission of HIV and other STDs.

Increases in condom-use consistency differ across racial and ethnic groups, however. Between 1988 and 1995, condom use among non-Hispanic white and non-Hispanic black teenagers increased by 31% and 17%, respectively. However, condom use among sexually active Hispanic male adolescents did not increase significantly over this seven-year period and, at 59% in 1995, remains well below the levels recorded among black and white teenage males.

While Hispanic teenagers may be employing other methods to protect against pregnancy, their low level of consistent condom use and the small increase in this level over time is cause for concern, since they have the highest rates of HIV and STD infection among all racial and ethnic groups studied.21This suggests that current programs and policies need to be modified to better serve Hispanic youths.

Second, condom-related attitudes changed significantly between 1988 and 1995. Attitudes about condom ineffectiveness, male contraceptive responsibility, AIDS risk and AIDS worry all changed in a direction suggestive of lower levels of consistent condom use. While it is possible that AIDS concerns are lower because rates of condom use have increased, reduced media coverage of the AIDS epidemic—together with advances in the treatment of infected individuals—also might have convinced adolescents that there is less reason to be concerned about contracting HIV.

On the other hand, attitudes concerning partner appreciation, pleasure and condom-use embarrassment all changed in a direction suggestive of higher levels of consistent condom use. This could be the result of successful education and advertising campaigns, which may have lowered the stigma associated with the discussion, purchase and use of condoms.

Changes in attitudes toward pregnancy prevention and AIDS avoidance in a direction suggestive of lower levels of consistent condom use may be the result of a shift in the composition of the study sample between 1988 and 1995. The concern that previously may have led young males to use condoms now may lead them to choose abstinence instead; consequently, a higher proportion of sexually active youths now are those who are less likely to take precautions. Moreover, youths who are more cautious in their sexual behavior may be those for whom correlations between attitudes and behaviors are the strongest.

Lastly, there were significant changes in the effects of attitudes on condom-use consistency between 1988 and 1995. The most important of these was the change in the impact of condom-use embarrassment. In 1988, there was a significant negative relationship between condom-use embarrassment and consistent condom use; by 1995, this relationship was no longer negative and no longer statistically significant. Thus, embarrassment about using condoms is less likely to be an obstacle than it was in the past.

Overall, our findings indicate that changes in attitudes toward condoms do not explain most of the large increase in condom-use consistency observed between 1988 and 1995. Only the three personal consequence attitudes changed significantly in a direction that predicts higher levels of consistent condom use. However, for two of these measures, the relationship between the attitude and condom-use consistency changed in a direction predicting decreased condom use. Only in the case of condom-use embarrassment did the relationship between attitudes and condom use change in a direction that predicts more consistent condom use over time.

If changes in condom-related attitudes and in the relationship between condom-related attitudes and condom use do not adequately explain the increase in consistent condom use, what might? One possibility is that condom use increased because of changes in the characteristics, attitudes or behaviors of the respondents' female partners. While we have no firsthand information on these women, we can surmise something about them from their partners. Males in the 1995 cohort were significantly more likely than those in the 1988 sample to report that their female partner would appreciate their use of a condom, and that she would not be upset if she knew her partner had a condom ready.

In addition, the males in the 1995 sample were less likely than those in the 1988 sample to indicate that they would feel embarrassed putting a condom on in front of their partner. Taken together, these findings suggest that the partners of adolescent men are increasingly likely to request or insist on the use of a condom during sexual intercourse. Perhaps those females who appreciate their partner's use of a condom are now saying no to sex without one.

Efforts to further increase the level of consistent condom use among adolescent males should focus on emphasizing that responsibility for discussing, purchasing and using contraceptives rests with both partners. Young males should continue to be educated regarding the risks and consequences of contracting HIV and other STDs. Finally, greater efforts must be made to target programs toward Hispanic youths, both males and females.