Preadolescents' perceptions of AIDS before and after Earvin Magic Johnson's announcement.

Compared sixth gratkn' AIDS concerns befon and after Magic Johnson an nounced thal he tested HIV-positive. Examined perceptions of self/other vul nerability (optimistic bias) using questionnaires (Study A) and inlerviews (Study B ). Also e.xamined AIDS worries mentioned in open-eruhd interviews (Study C ). Studies A and B revealed a substantial degree of optimistic bias. Study C showed that AIDS is a p~ concern of school-age students, mentioned wuhr frt!e or contat-c1«d condilions by about 50%. Then was little tkuctable effect of the publicity; the sok befon-after difference in thne stud~s was one incnase in self- and other-vldnerabiliry to AIDS-an incnase attributable, with marginal significance, only to girls. Discussed implicmions for health promotion and dijfennces between assessment approaches.

The experiences and comments of well-known musicians, athletes, and actors might be expected to have a significant impact on young people, given the well-documented finding that adolescents obtain more of their infonnation about AIDS from television than from parents, teachers, or any other source (Fassler et al., 1990;Henggeler, Melton, & Rodrigue, 1992;Lindauer, Schvaneveldt, & Young, 1993;McElreath & Roberts, 1992). On a more general level, media m~ages have been shown to influence children's health-promoting and hcalthrisking behaviors (Siegel, 1993). 1bc present paper focuses on a highly salient type of media message: celebrity disclosure. The goal was to compare students' concepts and concerns about AIDS before and after Magic Johnson •s announcement that he was HIV-positive. 1bc major research questions were whether this announcement would increase (a) estimates of the likelihood that people in general would contract AIDS, (b) perceptions of personal vulnerability, or (c) frequency of identifying AIDS as a cUITCnt concern or worry.

OVERVIEW
In three separate studies, sixth-grade students' perceptions of AIDS were assessed before and after the announcement that Magic Johnson had tested positive for HIV. The panicipants ranged in age from 10 to 13 years. All were attending public schools in Southern California. and most were Caucasian. Study A used a group-administered questionnaire asking students to indicate the likelihood that they themselves and the likelihood that people like them would be affected by specified health and environmental events, while Study B asked similar questions using a private interview format. In Study C, students provided open-ended responses rather than ratings. At the beginning and again at the end of an extensive interview involving multidimensional judgments of risk and What arc the chances this will happen to a typical person r Each event was presented twice on a single page. once for the SV and once for the OV question. Judgment! were made on IO-point scales that ranged from will not happen to will happen. The anchors on either side of the midpoint were mighl not happen and might happen.
The 19 problems were arranged in one of four random orders and presented in response booklets. Order of self· and other-judgments was fixed for each student but counterbalanced within each classroom such that bal{ of the participants always made OV judgments first and half always estimated SV first. In other words, students were randomly assigned to one of four problem orders, and approximately half in each order judged OV and then SV, while the other half judged SV first and then OV.
Administered in school classes, the questionnaire took about 15-20 minutes to complete. A trained graduate student gave instructions while another research staff member heJped with materials and answered questions. Teachers and other school personnel did not participate in the study. The students were assured that their responses were confidential and anonymous, and they were asked not to put their names on their questionnaires. The list of events was read aloud before the questionnaires were distributed, and students were encouraged to ask questions if any of the tenns were unfamiliar or unknown. Standard definitions were provided on request, but these requests were rare. The rating scale was presented as a type of ruler with I 0 marked sections but no numbers. For practice, the students began by indicating their chances of a relatively common event (i.e. , being at a ball game on a day the home team wins) and a relatively uncommon event (finding a $5 bill on the way home from school).
Responses were multiplied by 10 to create 100-point scales. The data were then analyzed in a 2 x 2 x 2 mixed analysis of variance design. There was one within-subject factor, optimiStic bias (OB), which contrasted estimates of OV and SV. The other two factors were between subjects: Gender and Time (before announcement or Tune 1 vs. after announcement or Time 2).
The means by gender and time are presented in Table I. There was a substantial optimistic bias effect, F(l, 240) = 274.33, p < .001 , with OV estimates almost twice those for SV. There were no differences between Time I and Time 2, indicating that Magic Johnso~ 's announcement bad no discernible sity of California, Irvine on August 18, 2016 effect on early adolescents• perceptions of AIDS when measured in this manner.
Finally, neither the main effect of gender nor its interactions approached significance.

Participants
The first phase, conducted between May and November 1991, involved 24 girls and 17 boys. The second 'phase, conducted in November 1991 immediately after the announcement, included a new sample of 15 girls and 17 boys. Prior to their participation, all students had returned signed parental consent forms and given their verbal assent as well. None of these students had participated in Study A.
An interview protocol that focused explicitly on selfand other-vulnerability was developed to para)lcl the questionnaire used in Study A. Using the same four event orders developed for Study A, we randomly assigned each student to one list for SY and to a different list for OV. The full set of events was rated along one dimension and then along the other dimension, with the SV-OV order counterbalanced.
These 20.minute interviews were conducted in a private setting at school such as the nurse's office. The students were told that the interviewer was interested in their opinions about different things that happen to people and to the environment. 1bcy were reassured that there were no right or wrong answers and that their responses would be confidential.
A ring with a 2-inch human figure attached to it was placed in the middle of a rcxt such that it could slide in either direction. The figure was portrayed as representing either a typical person (OV) or the student (SV). For selfvulnerability, the students were asked to ••pretend that this person is you. When l read one of the cards to you, tell me how likely it is that it will happen to you sometime during your life, any time from now on." For other-vulnerability, students were asked to .. pretend that this figure is someone else just like youjust a typical person your age. When I read one of the cards, think of other people like you in the United States and tell me how likely you think it is that it will happen to a typical person sometime during his or her life, any time from now on." The structured interview protocol was designed for a larger study of young people's health infonnation processing. The major components involved perceptions of the 19 common health and environmental risks described above. This study diff ercd from the others in that four dimensions were a.sscsscd: seriousness or severity, population prevalence, personal vulnerability, and perceived controllability over the event or its consequences. As noted above, these dimensional findings will be reported in a subsequent publication. The focus here is on a more qualitative component. At the beginning and the end of the session, open-ended questions were used to elicit students' primary worries or concerns, both in general and as related specifi~ly to (a) health and (b) the environment. The students were asked to identify the things that they worry or "think bard" about. The latter phrasing was included because, during pretesting, several students asserted that they just did not worry.
Once again, interviews were conducted in private. One of five female graduate students conducted each interview, working with an undergraduate observer who assisted with materials and response recording. Interviewers and observers received systematic training, including role-play exercises and observations of ongoing interviews, and foJlowed a set of explicit guidelines. The interviews took approximately 45 minutes to 1 hour, and all were audiotaped.
Students were assured that everything they said was confidential and that there were no right or wrong answers. Given the focus on worries and concerns, special care was taken to prevent iatrogenic (interview-produced) distress. The interview concluded with a positive mood induction phase during which the student was asked to discuss .. good things that happen to people-things that make you and others your age f~l good." The interviewers were trained not only to conduct the interview in a sensitive manner but also to monitor for any signs of anxiety or distress so that the session could be discontinued. No student appeared distressed by the procedures, and no session was truncated.

Coding
The open-ended responses were transcribed from the audiotapes, and the transcriptions were checked and corrected as needed by a second coder. Once these verbatim transcriptions were obtained, three raters counted the number of students who mentioned AIDS during any of these open-ended components. A stringent criterion was applied such that only those responses that explicitly mentioned AIDS were included. Worries about deadly viruses or incurable illnesses were not counted because such comments could refer to other catastrophic diseases. Intcrjudge agreement was 100%.

DISCUSSION
Three noteworthy conclusions can be drawn from the studies reported here. The first finding is that even sixth-grade children show a high degree of concern about AIDS, with approximately one in four mentioning AIDS spontaneously when asked to describe current worries. This figure increases to 50% when those who mention it at the beginning of the interview are combined with those mentioning it at the end, after they bad evaluated AIDS as one of 19 health and environmental problems.
A second conclusion is that, like adults, preadolescents evidence a substantial degree of what has been called optimistic bias when gauging their risk of contracting AIDS . Whether assessed through individual interviews or group questionnaires, boys and girls 'rated unknown but similar others as substantially 11lQfe likely than themselves to contract AIDS . This finding is consistent with previous studies demonstrating that older adolescents and adults display what has been labeled unrealistic optimism across a broad array of risks and hazards, including AIDS (Gladis, Michela, Walter, &. Vaughan, 1992;Kirscht et al., 1966;Ncmcroff, Brinkman, & Woodward, in press;Weinstein, 1984).
A third conclusion is that Magic Johnson's announcement that be had tested positive for HIV, while it galvanized public and media interest, had at most a modest impact on students' risk perceptions. Three diverse methodologies were used to probe perceptions about AIDS before and after the announcement. During private interviews about perceived vulnerability to AIDS, both self-and other-risk perceptions of sixth-grade girls were higher after than before the announcement. There were increases in vulnerability assessments for both the self and an unknown other, but no change in the level of optimistic bias. This difference was not found for boys, nor did differences emerge for either gender during the group survey of a comparable sample of sixth-grade students. Analogously, no significant difference was found in a third study that included nondirectivc questions about worries and concerns, even though there was a 12% increase in the number of youngsters spontaneOUSly mentioning AIDS after the announcement.
Both the assessment context and the procedures must be considered when interpreting these findings. None of the studies focused explicitly on AIDS. To the contrary, AIDS was embedded in a set of 18 other health and environmental problems for Studies A and· B, and in Study C, the questions about worries were open-ended and nondircctivc. Thus, the most valid conclusion is that the effects of the announcement on students' risk perceptions, when considered within a broad array of everyday and catastrophic concerns, were quite modest.
Ineluctably, there are limitations and uncertainties when studies are launched in response to an unexpected event rather than following extensive planning and pretesting. Ideally, we would have interviewed individual students School contacts, subject recruitment, and data collection efforts needed to be reestablished for the other two studies and, consequently, more time (i.e. , 1 to 2 months) elapsed between the announcement and the Time 2 measures. In subsequent studies of the impact of such salient events, it would be enlightening to collect information during two or more postevent intervals so that the effects of fading exposure and salience can be examined. 1be need for such information is underscored by a study of risk perceptions following a major earthquake in California (Burger & Palmer, 1992). Optimistic bias concerning natural disasters disappeared immediately following the earthquake but returned within 3 months.
Also noteworthy was the specificity of this effect 1'berc were no analogous effects on perceived vulnerability to unrelated baz.ards such as getting mugged or developing a drinking problem.
It is unclear why the aftef-announcement interview ratings were higher for girls but not for boys. Previous studies using diverse methodologies have demonstrated that females tend to acknowledge health concerns more readily than do maJes (Brown & Fritz, 1988;DiClcmcnte, Zorn, & Tcmosbok, 1987;Gochman & Saucier, 1982;Kirscht et al., 1966). If this oft-found pattern reflects differential concern rather than merely differential willingness to report such concerns, it would follow that girls may be more receptive than their male counterparts to new, risk-relevant infonnation, or perhaps more likely to integrate salient events into their evolving cognitive schemas. This gender difference may have been enhanced by the fact that the graduate student interviewers were female. Consistent with this Study B pattern is the fact that, during the open-ended elicitation of worries in Study C , there was a 19% increase in the proportion of girls who mentioned AIDS after the announcement, in contrast to a 5% increase for boys.
These gender differences should be interpreted with caution, however. In Study C, the gender effect failed to reach statistical significance. In Study B, the before-after difference is largely attributable to the fact that the Time 1 assessments of girls were substantially lower than those of boys, not to any gender difference in Time 2 levels. This series of studies focused exclusively on the cognitive realm; behaviors and intentions were not assessed. Given that it is often easier to modify cognitions than actions (Whalen & Kliewer, in press), it seems reasonable to conclude that the announcement has not had a significant impact in the behavioral realm.
Indeed, a failure to find after-announcement changes in the sexual practices or intentions of high school students buttresses this assumption (Who's Who Among American High School Students, 1992). The modest findings reported here should not be interpreted, however, as evidence against the potential significance of affectively laden, real-life events or the educational impact of hero figures. The present study assessed only undirected ripple effects and reminded us that we cannot assume beneficial outcomes no matter how poignant the naturally occurring intervention. The impact of systematically harnessing and in-