The association between Magic Johnson's HIV serostatus disclosure and condom use in at-risk respondents.

Two waves of data from the National AIDS Behavioral Surveys, one collected prior to Magic Johnson's disclosure that he was HIV positive and one collected after, were used to address whether the disclosure was related to increases in condom use in a representative sample of respondents with a risk factor for HIV. Results indicated that Blacks, Hispanics, those with fewer than 12 years of education, and unmarried respondents were more likely to report being influenced by Magic Johnson's announcement. In addition, even when sociodemographic variables related to change in condom use were controlled, those respondents who reported being influenced by Magic Johnson's announcement were more likely to increase their condom use.

When Magic Johnson announced that he was HIV positive in November 1991, public health professionals hoped that his celebrity status would help increase public awareness of IRV and AIDS and promote the practice of safer sex behaviors. Although the announcement may have had an impact at some level, no researchers to date have looked at the change in HIV-prevention behavior in a representative sample of participants at risk for HIV (Kalichman, 1994). We addressed three important sets of questions that remain to be answered regarding the reported impact of Johnson's disclosure. First, what are the differences between the people who reported that they were influenced by Magic Johnson's announcement and those who were not? For example, consistent with the perspective that the influence of a role model is increased by similarity to the target person (Bandura, 1986), are younger Black males, who may view Magic Johnson as a role model, more likely to report being influenced than other ethnic or age groups? Are males more likely than females to be influenced by Johnson's announcement? Most researchers have focused on a single ethnic or age category, so the differences between strata could not be investigated (e.g., Boccher-Lattimore, Jemmott, & Jemmott, 1994; Sigelman, Miller, & Derenowski, 1993; Whalen, Henker, O'Neil, Hollingshead, Holman, & Moore, 1994; Zimet, Lazebnik, DiClemente, Anglin, Williams, & Ellick, 1993). Those researchers who have made comparisons have found that ethnicity, gender, education, and knowing someone with AIDS are associated with whether a person reports being influenced by Magic Johnson's announcement (Gleghorn, Kilbourn, Celentano, & Jemmott, 1993; Kalichman, 1994; Kalichman & Hunter, 1992; Kalichman, Russell, Hunter, & Sarwer, 1993; Langer, Zimmerman, Hendershot, & Singh, 1992).

Second, did Magic Johnson's disclosure contribute to change in behavior within individuals? With one exception, researchers have looked at mean levels of behavior before and after the announcement in independent groups, but have not assessed the same individuals at both points to allow intraindividual assessments of change (e.g., Boekeloo, Schiavo, Rabin, Jordan, & Matthews, 1993; Gleghorn et al., 1993; Kalichman & Hunter, 1992; Kalichman et al., 1993; Tesoriero, Sorin, Burrows, & LaChance-McCullough, 1995; Langer et al., 1992; Zimet et al., 1993). In one study in which researchers did examine individual change pre- to post-announcement, Boccher-Lattimore and colleagues measured Black adolescents' beliefs about condoms and intentions to use condoms, recruiting students from public junior high schools (Boccher-Lattimore et al., 1994). The students completed questionnaires five months prior to Magic Johnson's announcement and then again one month after the announcement. After the announcement, the students reported stronger intentions to use condoms in the next three months, increased self-efficacy to negotiate condom use, and increased perceived access to condoms. The results provided evidence that Black adolescents' attitudes and behavioral intentions were associated with Magic Johnson's announcement but did not address whether participants of different races, ethnicities, or ages were more or less likely to report being influenced in the practice of HIV-prevention behaviors such as condom use.

A third question that remains to be answered is whether the announcement was associated with an increase in safer sex behavior. In three studies researchers have measured change in reported behavior in convenience samples attending sexually transmitted disease (STD) clinics. Gleghorn et al. (1993) found that the proportion of patients using condoms at least some of the time increased significantly in those interviewed after the announcement. In another study of STD clinic patients, respondents surveyed after the announcement reported fewer one-night stands and fewer instances of three or more partners during the previous three months (Boekeloo et al., 1993) than respondents surveyed before Magic Johnson's announcement. Langer et al. (1992) found that approximately 70% of their sample reported a change in sexual behavior (e.g., using condoms, becoming monogamous or abstinent) in response to Magic Johnson's disclosure. Again, however, these researchers did not look at within-individual change in behavior.

As part of a longitudinal study of HIV-related behaviors and attitudes, we tested the hypothesis that increase in condom use was related to respondents' reports of being influenced by Magic Johnson's announcement. When Magic Johnson made his disclosure in November 1991, we were fortunate to be between waves of a longitudinal study. Thus, we were able to assess the relation between the announcement and changes in variables that we measured in both waves of the study. Our main hypothesis was that those respondents who reported being influenced by Magic Johnson's announcement would be more likely to increase their condom use than those respondents who said they were not influenced by the announcement. We examined the unique relation between reported influence and increase in condom use while statistically controlling other sociodemographic variables related to condom use. In addition, we examined the demographic and attitudinal correlates of reported influence.

Methods

Sample and Procedures

The data for the current analyses came from the High Risk Cities sample of the National AIDS Behavioral Survey (NABS), a longitudinal telephone survey of sexual behaviors and attitudes. The High Risk Cities sample consists of 18- to 49-year-old respondents from 23 U.S. cities that had relatively high rates of AIDS cases at baseline (in 1990): Atlanta, Baltimore, Boston, Chicago, Cleveland, Dallas, Fort Worth, Detroit, Houston, Los Angeles, Miami, New Orleans, New York, Philadelphia (including Camden, NJ), Pittsburgh, San Antonio, San Diego, San Francisco, Oakland, San Jose, East St. Louis, Tampa (including St. Petersburg), and Washington, DC. Participants were chosen by random digit dialing. During the first wave of the survey (June 1990 to February 1991), 8,263 people 18 to 75 years of age were interviewed. The second wave of the survey was conducted from January to September 1992. As noted earlier, Magic Johnson disclosed his HIV-positive serostatus in November 1991. Of those who volunteered to be reinterviewed, 66.4% (3,728/5,614) of Wave I respondents 18 to 49 years of age in the high-risk cities sample were interviewed in Wave II.

Risk status. At Wave I, a respondent was classified as at risk for HIV if he or she reported injection drug use in the past five years, had more than one sexual partner in the past year, received a blood transfusion between 1978 and 1985, was ever treated for hemophilia, or had a risky main sex partner. The main sex partner was defined as the respondent's most frequent sexual partner during the past 12 months, and the main sex partner was considered risky if he or she had sex with anyone besides the respondent, had ever injected drugs, was a hemophiliac, had received a transfusion after 1977 but before 1985, or was HIV positive. At Wave II a respondent was classified as at risk if he or she reported injection drug use in the last 12 months, had multiple partners in the past 10 years, had a main sex partner who had sex with anyone besides the respondent, had ever injected drugs, or was HIV positive. Participants who had a risk factor for AIDS at Wave I or Wave II were asked detailed sexual behavior questions, including items concerning condom use. Questions related to Magic Johnson's announcement were asked only of heterosexuals with a risk factor at Wave I or Wave II. In other words, respondents who were considered at risk in Wave I were asked the detailed sexual behavior questions and the Magic Johnson questions in Wave II regardless of their risk status at Wave II. Respondents who were not considered at risk in Wave I but were considered at risk in Wave II were also asked the detailed sexual behavior questions and the Magic Johnson questions in Wave II. (For a discussion of the patterns of change in risk status for respondents in the NABS, see Catania et al., 1995.)

Weighting The data reported here were weighted to adjust for unequal probabilities of selection and response. The original Wave I weights were adjusted for attrition using all data in the Wave I data set and then post-stratified to match population figures from the Current Population Survey (CPS) of 1989 (the year in which the original sample was drawn). The data analyzed in this report are thus representative of individuals living in the High Risk Cities who had a risk factor at Wave I of the study. In addition, each analysis (and thus each subsample) was reweighted to reflect the 1989 CPS.

Measures

Standard demographic measures were used to assess the demographic characteristics at baseline. They were categorized as follows: race/ethnicity (White, Black, or Hispanic), income ([is less than] $10,000, $10,000 - $20,000, $20,001 - $40,000, and over $40,000), education (less than 12 years, 12 years, more than 12 years), marital status (married, cohabiting, separated/divorced/widowed, never married), age (18-29, 30-39, 40-49), and gender. Respondents were also asked if they had ever known anyone with AIDS.

Condom use. An overall condom use score was calculated based on the number of times the respondent had vaginal or anal intercourse in the past six months and the number of times a condom was used during anal or vaginal intercourse. Condom use was not normally distributed. Therefore, the continuous condom use variable was categorized as never (0% of the time), sometimes (1% to 49% of the time), most times (50% to 99% of the time), and always (100% of the time). We calculated a score to indicate change in condom use from Wave I to Wave II for each respondent, and we defined a change in frequency of condom use as a change in category. For example, a person who reported "sometimes" using condoms in the Wave I survey, but then reported using condoms "most times" in Wave II was categorized as increasing condom use. To test our hypothesis that Magic Johnson's announcement was related to increased condom use, we dichotomized change in condom use into "increased" versus "decreased/stayed the same." Self-reported condom use has been found to be a valid measure of condom use in a variety of samples, as indicated by STD and HIV infection rates (see Catania, Dolcini, & Coates, 1992).

Influence of Magic Johnson's announcement. Respondents who reported vaginal or anal intercourse with other-gender partners in the past six months were asked two questions regarding Magic Johnson's disclosure that he was HIV positive. First, participants were asked, "Are you aware of Magic Johnson's announcement that he has the AIDS virus?" Then, if they were aware of the announcement, participants were asked, "In deciding to use condoms during the last six months, did his announcement influence you to use condoms a lot, a little, or not at all?" For the analyses, responses to the latter question were grouped into "a lot/a little" versus "not at all." In this way we could compare those who reported any influence of the announcement to those who reported that their condom use was not influenced at all. The questions regarding Magic Johnson's announcement were asked after condom use had been measured, so it is unlikely that respondents were adjusting their condom use responses to be consistent with their answers to the influence question. Because the question concerning Magic Johnson's disclosure was framed in terms of influence on condom use in the past six months, only those participants who were interviewed at least six months after the announcement were included in the analyses of condom use. The restricted sample differed from the full sample on three variables. Respondents in the restricted sample were less likely to make more than $10,000 a year, were less likely to be married, and were more likely to be Black than were respondents in the full sample.

Results

Statistical Analysis

Chi-square analyses were used to assess bivariate relationships. Multivariate models were tested using hierarchical logistic regression. Two logistic models were assessed: one with whether the announcement influenced condom use as the dependent variable ("lot/little" versus "not at all"), and the other with change in condom use from Wave I to Wave II ("increased" versus "decreased/stayed the same") as the dependent variable. The independent variables for the initial models (ethnicity, age, gender, education, income, knowing someone with AIDS, and marital status) were chosen based on previous findings of the correlates of Magic Johnson's announcement (e.g., Glegborn et al., 1993; Kalichman, 1994; Kalichman & Hunter, 1992; Kalichman et al., 1993; Langer et al., 1992).

The final logistic models were attained by a four-step process. First, the independent variables were entered stepwise. Second, interactions were tested individually on the final step of the regression. Then a model was run in which only significant predictors were retained (p [is less than] .05) to achieve fit of the model and to allow interpretation of the parameters. On the last step, the standard errors were adjusted to account for the complex survey design. Unless otherwise noted, all results reported here are weighted.

The majority of respondents had heard of Magic Johnson's disclosure. Of the 2,086 respondents (unweighted N) who were asked, 99.2% indicated that they were aware of Magic Johnson's announcement that he was HIV positive. Of those interviewed at least six months after the announcement (N = 619), 24.8% indicated that their condom use was influenced a lot by the announcement, 12.6% indicated that their condom use was influenced a little, 60.5% reported that their condom use was not influenced at all, and 2.1% indicated that they did not use condoms. Descriptive variables for each subsample appear in Table 1.

Table 1
Distributions of Variables for Each Subsample

Variable             N = 2070(a)     N = 619(b)    N = 236(c)
Gender
 Male                   54.3            55.5          60.4
 Female                 45.7            44.5          39.6
Age
 18-29                  52.7            57.4          59.8
 30-39                  32.3            29            25.3
 40-49                  15              13.7          14.9
Ethnicity
 Black                  27.5            33.4          34.8
 Hispanic               12.4            12.3           8.4
 White                  56.8            49.1          50.4
 Other                   3.3              5.          36.4
Education
 > 12 years             14              15.1          17.8
 12 years               24.5            27.9          26.4
 < 12 years             61.5            57            55.8
Income
 < $10,000              20              25            28.5
 $10,000 - $20,000      22              22.5          21.3
 $20,001 - $40,000      34.6            30.9          30.3
 Over $40,000           23.5            21.6          19.9
Marital Status
 Married                35.2            29.1          15.8
 Cohabiting             14.5            15.6          15.2
 Separated/
  Divorced/Widowed      15              16.2          20.7
 Never married          35.4            39.1          48.3
Know someone
  with AIDS?
 Yes                    27.4            28.7          32.9
 No                     72.6            71.3          67.1
Magic Johnson
  announcement
  influence condom
  use?
 Yes                    35              38.2          39.6
 No                     65              61.8          60.4

(a) N = 2 070: The number of respondents who were aware of Johnson's announcement out of the 2,086 respondents who were asked

(b) N = 619: The number of respondents who were interviewed at least six months after the announcement

(c) N = 236: The number of respondents who had condom use scores at both Wave I and Wave II

Relationship of demographic variables to influence variable. To address the question of who was influenced by Magic Johnson's announcement, we used a logistic regression model with participants' responses to the question of how much their condom use was influenced by the announcement dichotomized into "Lot/Little" versus "Not at all" as the dependent variable (i.e., some effect of the announcement versus no effect of the announcement). Independent variables in the initial model were ethnicity, age, gender, education, income, knowing someone with AIDS, and marital status. To control for any effect of time, the number of months between the announcement and the interview was included in the model. On the final step, interactions between number of months since the announcement with ethnicity, education, and gender were tested in separate models. In the final model, only significant independent variables (p [is less than] .05) were retained. Results are in Table 2.

Table 2 Association of Demographic Variables with Response to Whether Magic Johnson's Announcement Influenced Condom Use.

Variable (Reference      Odds Ratio     95% Confidence Interval
  Category)
Ethnicity (White)
 Black                      2.94          1.79 - 4.82(**)
 Hispanic                   2.33          1.21 - 4.47(**)
Education (> 12 years)
 < 12 years                 3.08          1.56 - 6.13
 12 years                   1.60           .96 - 2.67
Marital status (Married)
 Cohabiting                 1.58           .76 - 3.30
 Separated, divorced,
   or widowed               1.98           1.00 - 3.92
 Never married              3.89           2.22 - 6.80

Note: N = 577. N for the logistic regression is slightly lower than the N = 619 because of missing data.

(*) p < .05 (**) p [is less than or equal to] .01

Ethnicity, education, and marital status at baseline were significantly related to whether respondents said their condom use was influenced by Magic Johnson's announcement. Compared to Whites, greater proportions of Blacks and Hispanics reported that their condom use was influenced by the announcement. A significantly larger percentage of respondents with fewer than 12 years of education reported being influenced by the announcement compared to respondents with more than 12 years of education. A significantly higher proportion of respondents who were separated, divorced, widowed, or never married indicated that they were influenced by the announcement. None of the two-way interactions between demographics and number of months since the announcement was significant in the final model.

Correlates of increased condom use. We were able to assess change in condom use for 236 at-risk respondents who had condom use scores at both Wave I and Wave II. The respondents in the sample with condom change scores (N = 236) differed from the respondents in the larger sample (N = 619) in that they were somewhat more likely to be single (separated, divorced, widowed, or never married).

Almost one quarter of the respondents (24%) reported increased condom use from Wave I to Wave II. Increased condom use was significantly related to the reported influence of Magic Johnson's announcement on condom use (RR = 1.81, 95% CI = 1.09 to 2.99). Thus, those individuals who said that Magic Johnson's announcement influenced their condom use "a lot" or "a little" were more likely to report increased condom use from Wave I to Wave II of the survey than respondents who reported no influence of the announcement on their condom use.

To determine whether Magic Johnson's announcement was related to condom use when other correlates of increased condom use were controlled, we used a logistic regression with change in condom use as the dependent variable ("increased" versus "decreased"/"stayed the same"). In addition to the influence question, other independent variables were ethnic group, age, gender, education, income, knowing someone with AIDS, months since the Magic Johnson announcement, and marital status. Interactions of influence with ethnicity, gender, month of interview, and education were entered on the last step in separate models. Only those independent variables that were statistically significant were retained in the final model. Results appear in Table 3.

Table 3 Association of Magic Johnson's Announcement and Demographic Characteristics with Change in Condom Use

Variables (Reference      Odds Ratio     95% Confidence Interval
  category)

Ethnicity (White)
 Black                      3.29          1.37 - 7.93(**)
 Hispanic                   6.36          1.73 - 23.37(**)
Education (> 12 years)
 < 12 years                 2.96          (see interaction below)
 12 years                   1.55          (see interaction below)
Income (< $10,000/year)
 $10-20,000                 1.06            .28-3.99
 $20,001 - 40,000           3.22            .72-14.29
 Over $40,000               6.09           1.16 - 32.08
Magic Johnson
  announcement
Influence condom
  use? (No)                 1.06           (see interaction below)
Marital status
  (Married)
 Cohabiting                  .56            .10-3.05
 Separated, divorced,
  widowed                   3.83           1.05 - 13.95(*)
 Never married              3.54           1.06- 11.84(*)
Education x Influence
(Yes vs. No
  comparisons within
  level of education)
 < 12 years x yes           8.00           1.54 - 41.73(*)
 12 years x yes            13.83           1.31 - 146.12(*)
 >12 years x yes            1.06            .39 - 2.86

Note: N = 233. N for the logistic regression is slightly lower than the N = 236 because of missing data.

Results indicated that whether the announcement had an influence on condom use was dependent on the education of the respondent. Among respondents with 12 or fewer years of education, a greater proportion of those who said the announcement had an effect on their condom use increased their condom use compared to respondents who said the announcement had no such influence on them. The relationship between the influence question and change in condom use was not significant in those with more than 12 years of education. In addition, ethnic group, income, and marital status were significant correlates of increased condom use. Specifically, larger proportions of Blacks and Hispanics reported increased condom use from Wave I to Wave II when compared to Whites. Higher income was also associated with increased condom use. Proportionally more respondents making over $40,000 a year increased their condom use compared to respondents making less than $10,000 a year. Among respondents who were separated, divorced, widowed, or never married, higher percentages reported increased condom use when compared to married individuals.

Discussion

Following Magic Johnson's disclosure, there was a significant increase in the number of calls to the Centers for Disease Control and Prevention's (CDC's) AIDS hotline (CDC, 1992), an approximate doubling of antibody testing at anonymous testing sites (Gellert, Weismuller, Higgins, & Maxwell, 1992; Tesoriero, et al., 1995), changes in beliefs (e.g., Boccher-Lattimore et al., 1994; Gleghorn et al., 1993; Sigelman et al., 1993), knowledge (Gleghorn, et al. 1993; Kalichman & Hunter, 1992; Sigelman et al., 1993), and concerns regarding HIV and AIDS (Gleghorn et al., 1993; Kalichman & Hunter, 1992; Kalichman et al., 1993; Zimet et al., 1993), as well as an increase in some safer sex behaviors (Boekeloo et al., 1993; Langer et al., 1992). However, to our knowledge, ours is the first study to assess the impact of Magic Johnson's disclosure of his HIV serostatus on a representative sample of respondents who are at risk for HIV infection. It is one of the few studies in which the same individuals were interviewed pre- and post-announcement, enabling us to look at intraindividual change in behavior. In addition, we examined change in condom use, not just intentions to use condoms. Although self-reports of condom use cannot be easily verified, self-reported condom use is a valid index of actual condom use, as indicated by STD rates (Catania et al., 1992).

Approximately 37% of the respondents with a risk factor for HIV in our sample of residents of high-risk cities reported that their condom use was influenced by Magic Johnson's announcement. Those who reported being influenced were more likely to increase their condom use than those not influenced, and this effect remained significant even in the presence of other sociodemographic variables. Thus, we provide evidence that not only were people reporting that the announcement had an impact on their condom use, but also that those who reported that the announcement had an impact on their condom use actually increased their condom use.

Not all those who increased their condom use increased it to 100% use at Wave II. Of the 56 people who increased their condom use, 39% increased it to "always" or 100% condom use. An additional 13% of respondents with condom use scores at both waves were consistent "always" condom users. Thus, substantial numbers of at-risk individuals were not motivated to increase their condom use (147 of 233, 63%) or, if they did increase condom use, to use condoms every time they had sexual intercourse (34 of 56, 61%). However, even though all respondents who increased their condom use did not become "always" users, there is evidence that even a modest increase in condom use is reflected in a decreased HIV seroconversion (Ngugi et al., 1988; Shelton & Harris, 1990).

Consistent with Social Cognitive Theory, which states that the influence of a role model is increased by similarity to the target person (Bandura, 1986), we found that Blacks were more likely to report being influenced by Magic Johnson's announcement than were Whites. Others who have investigated the effect of Johnson's announcement found that Blacks reported a greater impact of the announcement than Whites (Kalichman & Hunter, 1992; Kalichman et al., 1993; Langer et al., 1992). Although Marin and Marin (1990) found that Hispanics rate sports figures as the least credible source of AIDS information, our data indicate that Hispanics also were more likely to report being influenced by the announcement than Whites. It may be that Hispanics in our sample identified with Magic Johnson's status as an ethnic minority rather than his status as a sports figure and, thus, were more likely to attend to the message he was conveying. We did not find complete support for a modeling perspective, however. Based on the idea that the more similar the respondent is to the target person, the more the respondent is likely to be affected by the target's message, men would be expected to be more influenced by Johnson's announcement. However, in our data, there was no indication that men were more likely than women to report being influenced by the announcement.

We found that other respondents who reported being affected were those with low levels of education. It is not immediately clear why there was a differential association at different levels of education; however, this finding is consistent with the results of Langer et al. (1992), who found that STD clinic patients with lower levels of education were more likely to report behavior change in response to Johnson's announcement. Finally, single respondents, who are more likely to be in the dating phase of life and, therefore, are most likely to accrue risk, also reported that their condom use was significantly influenced by Magic Johnson's announcement. It may be that single people recognized the parallels in their own risky behavior relative to Magic Johnson's and, thus, were more likely to be influenced by Johnson's disclosure.

When we examined correlates of increased condom use, we found that those who reported being influenced by Johnson's announcement were significantly more likely to report increased condom use, even when the effect of these other sociodemographic variables was controlled. This is despite the fact that we found that ethnicity, education, and marital status were associated with reported influence of Johnson's announcement. Thus, the effect of Johnson's announcement cannot be explained as simply a functions of ethnicity, education, income, or marital status.

If the relation between Magic Johnson's announcement and change in condom use is a causal one, there are several points in the process of behavior change at which the announcement may have had an impact. Most theories of behavior change posit several stages to the process of change in risky sexual practices (e.g., the AIDS Risk Reduction Model, Catania, Kegeles, & Coates, 1990; the Health Belief Model, Rosenstock, Strecher, & Becker, 1994). For example, the AIDS Risk Reduction Model (ARRM) illustrates three stages to behavior change: labeling of the behavior as risky, commitment to engage in safer sex behavior, and enactment of the behavior, which includes negotiating safer sex practices with the partner. Johnson's disclosure could have had an impact at each of these stages. At the labeling stage, the announcement may have decreased stereotypes of who is at risk and, thereby, increased perceptions of personal risk for HIV. The announcement may have increased commitment to use condoms--the second stage of the ARRM. In the press conference when he disclosed his HIV serostatus, Johnson explicitly stated that he wanted to influence people's intentions to engage in safer sex practices, and based on our data, it appears that he was successful. Finally, the announcement may have had an impact in the enactment of prevention behavior by serving as a point of discussion for couples negotiating the use of condoms. The extensive media coverage of Magic Johnson's announcement may have made it easier for one partner to bring up the topic of using condoms, thus facilitating the negotiation of condom use, something that would occur at the enactment stage of the AIDS Risk Reduction Model.

Several researchers who have studied the impact of Magic Johnson's disclosure found that after a brief initial effect of the announcement, the impact declined to baseline levels (Kalichman et al., 1993; Penner & Fritzsche, 1993). Thus, in his review of psychological and behavior changes following Magic Johnsons announcement, Kalichman (1994) concluded that the effects of the announcement were short lived. However, the majority of the studies of the effect of Magic Johnsons announcement were conducted within a few months of the announcement (e.g., Kalichman & Hunter, 1992; Gleghorn et al., 1993). Our data speak to longer term effects of Johnsons announcement, because the interviews analyzed here were done six months to a year after the announcement. Subsequent to his announcement, Johnson became very involved in AIDS-prevention and education efforts. Thus, people may have received multiple exposures that enhanced the impact of his initial disclosure and produced a maximum effect months later. However, there may also have simply been a lag between his initial announcement and apparent behavior change that reflects the time needed for the message to diffuse sufficiently to begin to alter community norms.

At a societal level, Magic Johnson's announcement may have served as a "critical moment." Critical moments are powerful, value-laden communications that supply people with symbolic frames for issues and, thus, contribute to the process by which changes occur in the way issues are perceived and evaluated by the general public (Pollock, 1994). Clearly, the majority of people were aware of Magic Johnson's disclosure, and a significant proportion of those who were at risk for HIV changed their behavior in response to the announcement. Although studies of the media coverage of Johnson's announcement indicate that time devoted to AIDS-related stories declined to near baseline after a short time (Kalichman, 1994), it may be that the effect of the announcement was further propagated through social diffusion after the media coverage ebbed. Magic Johnson's announcement may have had a deeper impact at the level of societal norms for sexual behaviors and discussion and awareness of sexual topics. By bringing a sensitive topic into the media and into the discussions of a greater proportion of the population, Johnson may have actually caused a shift in public consciousness about HIV and AIDS. Perhaps as a result of Magic Johnson coming forward with his HIV serostatus and the positive consequences that followed, more people will choose to reveal their HIV-positive status and share the way AIDS has affected their lives. As it becomes increasingly apparent that "it can happen to anybody" (Magic Johnson in his November 7, 1991, press conference), perhaps attitudes about HIV and AIDS will change, the importance of safer sex behaviors will be acknowledged, and acceptance for those infected with IRV will become the norm.

References

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Boccher-Lattimore, D., Jemmott, J., & Jemmott, L. (1994). The Magic Johnson effect: Celebrity disclosure of HIV status and longitudinal changes in adolescents' condom-use intentions. Paper presented at the Society of Behavioral Medicine, Boston, April.

Boekeloo, B., Schiavo, L., Rabin, D., Jordan, C., & Matthews, J. (1993). Sexual risk behaviors of STD clinic patients before and after Earvin "Magic" Johnson's HIV-infection announcement--Maryland, 1991-1992. Morbidity and Mortality Weekly Report, 42, 45-48

Catania, J., Binson, D., Dolcini, M., Stall, R., Choi, K-H., Pollack, L., Hudes, E., Canchola, J., Phillips, K., Moskowitz, J., & Coates, T. (1995). Risk factors for HIV and other sexually transmitted diseases and prevention practices among US heterosexual adults: Changes from 1990 to 1992. American Journal of Public Health, 85, 1492-1499.

Catania, J., Dolcini, M., & Coates, T. (1992). Catania and colleagues respond to Cohen and Dent [Letter to the editor]. American Journal of Public Health, 82, 1564-1565.

Catania, J., Kegeles, S., & Coates, T. (1990). Towards an understanding of risk behavior: An AIDS risk reduction model (ARRM). Health Education Quarterly, 17, 381-399.

Centers for Disease Control (CDC). (1992). Demand for information remains high following "Magic" Johnson announcement of HIV infection. CDC HIV/AIDS Prevention Newsletter, 3, 7-8.

Gellert, G., Weismuller, P., Higgins, K., & Maxwell, R. (1992). Disclosure of AIDS in celebrities. New England Journal of Medicine, 327, 1389.

Gleghorn, A., Kilbourn, K., Celentano, D., & Jemmott, J. (1993). "Magic" Johnson's HIV status: Impact on Black STD patients. Paper presented at the American Psychological Association, Toronto, Canada, August.

Kalichman, S. (1994). Magic Johnson and public attitudes toward AIDS: A review of empirical findings. AIDS Education and Prevention, 6, 542-557.

Kalichman, S., & Hunter, T. (1992). The disclosure of celebrity infection: Its effects on public attitudes. American Journal of Public Health, 82, 1374-1376.

Kalichman, S., Russell, R., Hunter, T., & Sarwer, D. (1993). Earvin "Magic" Johnson's HIV Serostatus Disclosure: Effects on men's perceptions of AIDS. Journal of Consulting and Clinical Psychology, 61, 887-891.

Langer, L., Zimmerman, R., Hendershot, E., & Singh, M. (1992). Effect of Magic Johnsons HIV status on HIV-related attitudes and behaviors of an STD clinic population. AIDS Education and Prevention, 4, 295-307.

Marin, G., & Marin, B. (1990). Perceived credibility of channels and sources of AIDS information among Hispanics. AIDS Education and Prevention, 2, 151-161.

Ngugi, E., Plummer, F., Simonsen, J., Cameron, D., Bosire, M., Waiyaki, P., Ronald, A., & Ndinya-Achola, J. (1988). Prevention of transmission of human immunodeficiency virus in Africa: Effectiveness of condom promotion and health education among prostitutes. Lancet, 2, 887-890.

Penner, L., & Fritzsche, B. (1993). Magic Johnson and reactions to people with AIDS: A natural experiment. Journal of Applied Social Psychology, 23, 1035-1050.

Pollock, P., III. (1994). Issues, values, and critical moments: Did "Magic" Johnson transform public opinion on AIDS? American Journal of Political Science, 38, 426-446.

Rosenstock, I., Strecher, V., & Becker, M. (1994). The health belief model and HIV risk behavior change. In R. DiClemente & J. Peterson (Eds.), Preventing AIDS: Theories and methods of behavioral interventions (pp. 5-24). New York: Plenum.

Shelton, J., & Harris, J. (1990). Role of the condom in combating global AIDS: The application of Sutton's Law to public health. In N. Alexander, H. Gabelnick, & J. M. Spieler (Eds.), Heterosexual transmission of AIDS (pp. 327-338). New York: Wiley-Liss.

Sigelman, C., Miller, A., & Derenowski, E. (1993). Do you believe in Magic? The impact of "Magic" Johnson on adolescents' AIDS knowledge and attitudes. AIDS Education and Prevention, 5, 153-161.

Tesoriero, J. M., Sorin, M. D., Burrows, K. A., & LaChance-McCullough, M. L. (1995). Harnessing the heightened public awareness of celebrity HIV disclosures: "Magic" and "Cookie" Johnson and HIV testing. AIDS Education and Prevention, 7, 232-250.

Whalen, C. K., Henker, B., O'Neil, R., Hollingshead, J., Holman, A., & Moore, B. (1994). Preadolescents' perceptions of AIDS before and after Earvin Magic Johnson's announcement. Journal of Pediatric Psychology, 19, 3-17.

Zimet, G. D., Lazebnik, R., DiClemente, R. J., Anglin, T. M., Williams, P., & Ellick, E. M. (1993). The relationship of Magic Johnson's announcement of HIV infection to the AIDS attitudes of junior high school students. The Journal of Sex Research, 30, 129-134.

Manuscript accepted June 11, 1996

Correspondence should be addressed to Judith Tedlie Moskowitz, 74 New Montgomery, Suite 502, San Francisco, CA 94105. Phone: [phone]. Fax: [phone]. E-mail: [e-mail].

Judith Tedlie Moskowitz Diane Binson Joseph A. Catania Division of General Internal Medicine, Department of Medicine, Center for AIDS Prevention Studies, University of California San Francisco