AIDS-RELATED ATTITUDES IN THE UNITED STATES: A PRELIMINARY CONCEPTUALIZATION

This paper offers a preliminary conceptualization of the psychological structure of AIDS-related attitudes among American adults and describes some of the social and psychological factors that affect those attitudes. Data were collected first from participants in focus groups in five U.S. cities and then from respondents in a national telephone survey. Two major psychological dimensions of attitudes were observed consistently. The first dimension, labeled COERCION/COMPASSION, includes judgments about the extent to which AIDS is viewed as highly contagious and requiring containment, through coercion if necessary. It also includes attributions of blame to people with AIDS. The second dimension, PRAGMATISM/MORALISM, includes judgments about the extent to which AIDS is viewed as controllable through research, public education and governmental sponsorship of behavior-change programs. The two attitude dimensions are not highly correlated. Regression analyses suggest that the two dimensions have different social and psychological antecedents and that these antecedents differ between White and Black Americans. Using the two factors, a tentative typology of responses to the AIDS epidemic is presented. Implications for AIDS education and policy are discussed

KEY WORDS: AIDS, public opinion about; homosexuality, attitudes toward; racial minorities.

As the United States enters the second decade of the AIDS epidemic, attitudes and beliefs concerning HIV-disease will play an increasingly important role in shaping societal response. Americans will be called upon to bear the epidemic's considerable economic costs and, increasingly, to respond individually to persons with AIDS in their schools, neighborhoods, workplaces, and families. AIDS-related initiatives and referenda will appear with greater frequency on electoral ballots, and AIDS-related policies will be included in candidates' campaign platforms. Consequently, understanding public reactions will be critically important for educating Americans about the epidemic, promoting enlightened public policy, and fostering compassion for persons infected with HIV.

Public attitudes surrounding AIDS are shaped by the complex characteristics of the epidemic. AIDS is a transmissible and, to date, lethal disease; personal reactions to it inevitably are influenced by concerns about individuals' own well-being and that of their loved ones. AIDS also is a highly stigmatized illness. Many persons perceived to be infected with HIV have been fired from their jobs, driven from their homes and socially isolated (Herek, 1990; Herek & Glunt, 1988). This stigma results both from the physical characteristics of AIDS (e.g., its negative effect on physical appearance and ability for social interaction; its communicability; its perceived lethality) and its psychosocial characteristics (i.e., its prevalence among such already-stigmatized groups as gay men, IV-drug users, Blacks, and Hispanics). In particular, attitudes toward gay men appear to exert an important influence on reactions to AIDS (e.g., Herek, 1990; Pryor, Reeder, & Vinacco, 1989; Stipp & Kerr, 1989;).

The research described in the present paper was designed to provide a better understanding of Americans' attitudes associated with the AIDS epidemic. Our goals were to identify the principal psychological dimensions along which attitudes toward persons with AIDS and AIDS-related public policies are organized and to assess the roles of various social and psychological factors in shaping them. In the course of the study, we developed a tentative typology for conceptualizing public reactions to AIDS. We also became aware of important racial differences in how the epidemic is perceived.

Data collection proceeded in two stages. First, exploratory research was conducted with questionnaire responses from a sample of adult focus group participants in five U.S. cities. Second, AIDS-related attitudes were assessed in a national telephone survey with a probability sample of American adults.

STUDY 1: THE FOCUS GROUP SAMPLE Method

During the spring and summer of 1988, 22 focus groups were conducted with 155 English-speaking adult participants in five U.S. cities: three groups each in Detroit, Michigan (n = 16), Houston, Texas (n = 29), Atlanta, Georgia (n = 29), and Lincoln, Nebraska (n = 27), and ten groups in New York, New York (n = 54). Overall, the racial composition of participants was divided nearly equally between Blacks (48% of males, 46% of females) and Whites (42% of males, 50% of females). The remaining 13 participants classified themselves as Hispanic, Asian, or of mixed ancestry.

Focus group participants were recruited locally through fliers and advertisements which offered them $20 (in Atlanta, Houston, Lincoln, and New York) or $25 (in Detroit) for participating in a two-hour discussion group as part of a psychology research project. AIDS was not mentioned in the recruitment announcements. When respondents called a telephone number listed in the advertisement, a research assistant explained that the focus groups were part of a national study in which researchers were trying to learn more about ordinary citizens' views on the AIDS epidemic. Only two respondents decided not to participate upon learning that the discussion topic would be AIDS; most callers seemed to become more interested in the project upon learning its topic. Participants were assigned to groups according to race and gender. Some groups were all-female, others all-male, and others mixed; some groups were all or predominantly Black, others all or predominantly White, and others racially mixed. Participants were directed to come to the focus group site (usually a local school or church building) at a specified time. When they arrived, they were greeted by a research assistant, paid in advance for their participation, offered refreshments, and instructed to complete a preliminary questionnaire that included knowledge and attitude items about AIDS.

The group discussion lasted approximately 90 minutes and covered such topics as the causes of AIDS, how AIDS is transmitted, sources of information about AIDS, and how the local community should respond to the epidemic. After the discussion, participants completed a second questionnaire that included a short form of the Attitudes Toward Gay Men (ATG) scale (Herek, 1988), demographic items, and additional scales not pertinent to the present paper. After they completed the final questionnaire, participants received a packet of informational materials concerning AIDS and were encouraged to ask questions of the researchers. Within two weeks after the group meeting, a follow-up thank-you letter was mailed to each participant, with a copy of the Surgeon General's report on AIDS (Koop, 1986) enclosed. Participants were encouraged to contact the authors at C.U.N.Y. if they desired additional information; no follow-up inquiries were received.

The present paper focuses on the questionnaire responses. A more detailed analysis of the content of the focus group discussions will be presented elsewhere.

Questionnaires

Knowledge About AIDS Transmission. Participants indicated the likelihood that each of 12 different behaviors could transmit AIDS. Responses were on a 5-point, Likert-type scale ranging from not at all likely to very likely. Of the 12, only two were behaviors generally recognized as capable of transmitting HIV ("sharing a hypodermic needle [`works']" and "during sex, having a man's semen [sperm/cum] enter one's body").

AIDS Policy Attitudes. Participants indicated their level of agreement with 13 different statements concerning policies that should be taken in response to AIDS. A 5-point, Likert-type scale was used, ranging from strongly disagree to strongly agree. The items were constructed to represent a diverse set of AIDS-related issues reported in print and electronic media since 1985. They were refined during various AIDS-related research projects conducted by the first author during the past several years.

Attitudes Toward Gay Men (ATG) Scale. The 5-item short form (Herek, 1988) was administered, with the same 5-point Likert-type scale as used for the AIDS Policy Attitudes items. The ATG is a highly reliable measure of heterosexuals' attitudes toward gay men; its validity has been well-documented (Herek, 1984a, 1987a, 1987b, 1988). High scores indicate higher levels of anti-gay prejudice.

Demographic Data. Information was obtained about participants' sex, age, race, frequency of attendance at religious services (never, once or twice a year, monthly, or weekly), and number of years of formal education.

Results

Factor Analysis and Item Analysis of Questionnaires

AIDS Knowledge. The Knowledge items fit logically into either of two groups: Knowledge about "intimate contact," i.e., the ways that HIV can be transmitted (2 items) and knowledge about "casual contact," i.e., the ways that it cannot be transmitted (10 items). This a priori categorization subsequently was supported by the results of a principal components factor analysis with two factors extracted. Needle-sharing and introduction of semen into the body loaded highly ( > .45) on one factor accounting for 14.3% of the variance, and nine of the other ten items loaded highly on a second factor accounting for 38.5% of the variance. Varimax and oblique rotations yielded essentially the same results. The one item not loading highly on either factor was "being a blood donor." Numerous national surveys have found that a significant minority of Americans believe this to be a possible route of HIV transmission (e.g., Dawson, 1988); its ambiguous status in the minds of some participants undoubtedly explains its lack of association with either factor. Additionally, the item "having sexual intercourse while wearing a condom" was dropped because it was considered to be somewhat ambiguous (since the likelihood of transmission would depend on whether the condom was used correctly). The remaining items were scored as two different scales: a 2-item Intimate Contact Knowledge scale, with high scores indicating accurate knowledge about the routes through which HIV can be transmitted (alpha = .54), and an 8-item Casual Contact Beliefs[1] scale, with high scores indicating overestimation of the risks posed by casual contact (alpha = .77).

AIDS Policy Attitudes. The attitude items were subjected to a series of principal components factor analyses with various numbers of factors extracted and with different types of rotations. A two-factor solution emerged as the most interpretable. As with the Knowledge items, varimax and oblique rotations yielded highly similar results.

The first factor, which accounted for 22.5% of the variance, included items advocating government distribution of condoms and clean needles for IV drugs, government sponsorship of AIDS research and education about safer sex techniques, and enactment of AIDS-related civil rights legislation. Informed by the focus group discussions, we hypothesized that endorsement of these items represents a willingness to suspend moral judgments about sexual behavior and drug use in the interests of preventing AIDS and thereby saving lives. Rejection of the items, we hypothesized, represents an unwillingness to "condone" such behaviors through government policies. We labeled this the PRAGMATISM/MORALISM factor.

The second factor, accounting for 16.6%,to of the variance, included items advocating quarantine, mandatory HIV-testing and public labeling of people with AIDS. We hypothesized that endorsement of these items is indicative of a perception of persons with AIDS as dangerous and requiring physical containment, through punitive measures if necessary; rejection of the items indicates a general view that people with AIDS are not dangerous and are deserving of compassion. We labeled this the COERCION/COMPASSION factor. Two 5-item additive scales were constructed from the items (see Table 1). The PRAGMATISM/MORALISM items were combined into an AIDS Pragmatism scale (alpha = .68) with high scores indicating endorsement of nonmoralistic pragmatic policies. The COERCION/COMPASSION items were combined into an AIDS Coercion scale (alpha = .66), with high scores indicating endorsement of coercive and punitive policies.

Attitudes toward Gay Men Scale. Internal consistency for the ATG scale was judged to be acceptably high (alpha = .83).

Variables Influencing AIDS-Related Attitudes: Exploratory Analyses

Intimate Contact Knowledge scores were positively correlated with AIDS-Pragmatism scores (r = .14, p < .05), whereas Casual Contact Beliefs scores were correlated with AIDS-Coercion scores (r = .44, p < .001). In other words, people with accurate knowledge about how AIDS is transmitted tended also to endorse the same sorts of measures advocated by public health authorities. Those with erroneous beliefs concerning how AIDS is not transmitted tended to endorse policies that impose restrictions on infected persons. Respondents could manifest both of these patterns simultaneously, since AIDS-Coercion scores were not significantly correlated with AIDS-Pragmatism scores (r = -.13). Nor were Intimate Contact Knowledge scores significantly correlated with Casual Contact Beliefs scores (r = -.10). Attitudes toward gay men (ATG scores) correlated strongly with AIDS-Coercion scores (r = .48, p < .001) and slightly with AIDS-Pragmatism scores (r = -.17, p < .05). People with negative attitudes toward gay men (high ATG scores) tended to endorse coercive policies, and were somewhat likely to reject nonmoralistic pragmatic policies. High scores on Casual Contact Beliefs also were associated with hostile attitudes toward gay men (r = .30, p < .001). The remaining interscale correlations were not statistically significant.

Separate regression analyses were conducted with AIDS-Coercion and AIDS-Pragmatism scores as dependent variables. The variables of sex and race (coded as dummy variables), education and age were entered on the first step of the equation. On the second step, multiplicative interaction terms for these demographic variables (e.g., race X gender) were entered. ATG and Casual Contact Beliefs scores were entered on the third step, along with religious attendance. In preliminary analyses, Intimate Contact Knowledge scores and two-way and three-way multiplicative interaction terms failed to explain a significant amount of variance in AIDS attitudes and so were excluded from the final regression equations (see Pedhazur, 1982, for an explanation of this approach; and Herek, 1987b, for an example).

The results for each equation were strikingly different. For AIDS-Coercion attitudes, three variables emerged as significant predictors: Casual Contact Beliefs (accounting for 10.4% of the variance), educational level (9.9% of the variance), and ATG scores (7.4% of the variance). For AIDS-Pragmatism attitudes, only religious attendance explained a significant amount (9.3%) of the variance (see Table 2).

Discussion

The results with the focus group sample suggested several important hypotheses for follow-up research. First, AIDS-related attitudes appear to be structured cognitively along two dimensions, tentatively labeled COERCION/COMPASSION and PRAGMATISM/MORALISM. Second, the finding that the factors are not highly correlated suggests that some people may manifest ambivalence in their AIDS-related attitudes, e.g., they may simultaneously support nonmoralistic pragmatic policies and coercive measures. We observed manifestations of this "do something" syndrome--a willingness to endorse any AIDS-related policy that promises action regardless of its likely efficacy (Schneider, 1987)--in the focus group discussions.

Third, the two dimensions are influenced by different variables. High AIDS-Coercion scores were associated with misinformation about casual contact, lower educational levels, and hostile attitudes toward gay men. High AIDS-Pragmatism scores, in contrast, were predicted only by infrequent attendance at religious services.

Finally, the content of the focus group discussions suggested to us that Blacks and Whites may have different views of the epidemic. In particular, we noted that Blacks expressed distrust of the government and of experts more often than did Whites. Blacks also more frequently expressed the belief that the AIDS epidemic is being used as an excuse to persecute racial minorities. Based on these observations, we hypothesized that AIDS-Coercion and AIDS-Pragmatism attitudes may have different social and psychological antecedents among Whites and Blacks.

These conclusions were tentative, given the relatively small size and nonrepresentative nature of the focus group sample. Their replicability and generalizability subsequently were assessed in a national telephone survey, described in the next section.

STUDY 2: NATIONAL TELEPHONE SURVEY

Between July 5 and August 10, 1988, telephone interviews were conducted by the staff of the New York City Study at the C.U.N.Y. Center for Social Research with a random sample of 1,078 English-speaking American adults. In all, 960 complete interviews were obtained (580 females, 364 males, 16 gender not coded). Of these, 784 respondents (82%) were White, 82 (9%) were Black, 31 (3%) were Hispanics, 34 (4%) were coded as "other," and 29 (3%) did not indicate their race. The sample was selected using Random Digit Dialing (RDD) techniques from the universe of all American households with telephones. Each respondent was required to be an English-speaking resident of the household at least 18 years of age. Once a telephone number was reached, the interviewer constructed a list of all eligible household members, from which a respondent was selected randomly. If that respondent was unavailable, callbacks were attempted until the interview was completed, the respondent refused to participate, or the study ended. The response rate was 47%.[2]

Questions Asked. From the 25-minute interview, three sets of items are relevant to the present paper (analyses of the remaining items will be reported elsewhere). First, respondents were asked their opinion about whether AIDS could be transmitted through each of 12 different routes; the items were modified versions of those administered earlier to focus groups.[3] Second, respondents were asked to indicate whether they agreed, disagreed, or were "in between" for 16 different AIDS-policy items. The items included those administered previously to the focus groups with some modifications of wording, as well as new items added to clarify the factor structure (see Table 3). Third, respondents were asked whether they agreed, disagreed, or were "in the middle" for the five items comprising the short form of the Attitudes Toward Gay Men (ATG) scale (Herek, 1988). Additionally, information was obtained about respondents' background, including sex, age, race, educational level, and frequency of attendance at religious services (on a 9-point scale ranging from never to several times each week).

Results

Factor Analysis of Items

AIDS Knowledge. As with the focus group sample, the items assessing knowledge about HIV described riskful forms of intimate contact as well as nonriskful casual contact. A variety of factor solutions and rotations were examined, with varying numbers of factors extracted. As with the earlier sample, a two-factor solution was the most interpretable; varimax and oblique rotations yielded similar results. The Casual Contact Beliefs factor accounted for 33.6% of the variance; the Intimate Contact Knowledge items accounted for an additional 10.4% of the variance. When the items were combined into scales, internal consistency for the Casual Contact Beliefs items was acceptably high (alpha = .84). The two-item Intimate Contact Knowledge scale, however, displayed low internal consistency (alpha = .11), apparently because one item ("receiving a blood transfusion") was somewhat ambiguous (since blood transfusions are now considered to be relatively safe from HIV-transmission). For this reason, the single item of "having sexual intercourse without using a condom" was used as an indicator of accurate knowledge about HIV transmission through intimate contact.

AIDS Policy Attitudes. As with the focus group data, a two-factor solution emerged as the most interpretable after a variety of factor solutions and rotations were examined. Individual item loadings were very similar with varimax and oblique rotations. The same two factors emerged: COERCION/COMPASSION (19% of the variance) and PRAGMATISM/MORALISM (11.6% of the variance). Two newly constructed items assessing attributions of blame to people with AIDS loaded on the COERCION/COMPASSION factor (see Table 3). Separate factor analyses with responses from Blacks and Whites yielded essentially the same factors. As with the focus groups, two 5-item scales were constructed, each of which displayed an acceptably high level of internal consistency: alpha = .70 for the AIDS Coercion scale, and .63 for the AIDS Pragmatism scale.

ATG items. The 5-item ATG scale displayed an acceptable level of internal consistency in this, its first telephone administration (alpha =.85).

Racial Differences

Racial comparisons were limited to White and Black respondents because of the small number of persons from other racial groups. Blacks and Whites differed in their willingness to trust the government and scientific authorities. Blacks more often agreed that the government is not telling the whole story about AIDS (67% of Blacks agreed, compared to 34% of Whites; chi-square = 29.84, p < .001). Whites were more likely to state that they believe scientists and doctors who say that AIDS is not spread through casual contact (71% of Whites agreed, compared to 58% of Blacks; chi-square = 9.84, p < .01). Blacks were somewhat more likely than Whites (51% versus 41%) to agree that the AIDS epidemic is being used to promote hatred of minority groups, although the difference was not statistically significant; chi-square = 2.99). A notable sex difference also was observed among Blacks: In estimating their own risk for contracting AIDS, Black men estimated their risk higher than Black women, White women, and White men; Black women estimated their risk lower than Black men, White men, and White women. Given the overrepresentation of African-Americans among the population of women with AIDS in the United States, the low level of perceived risk among Black women in our sample may point to an area requiring the attention of health educators. Our data do not permit a check of whether these self-estimates of risk were accurate, however.

Because Blacks were, on average, significantly younger than Whites (Blacks' mean age = 35.7, sd = 13; Whites' mean age = 42.3, sd = 15.8) and had lower educational levels (median educational level for Blacks was "high school graduate"; for Whites, it was "some college"), age and educational level were used as covariates in interracial comparisons for other variables.Comparisons using ANOVA revealed group differences on several variables: 1) Casual Contact Beliefs: Blacks were significantly more likely than Whites to overestimate the risks posed by casual contact (mean scores = 2.80 for Blacks and 1.38 for Whites, F (1,883) = 22.49, p < .001); 2) Attitudes Toward Gay Men: Blacks expressed more hostile attitudes toward gay men (mean ATG scores = 6.79 for Blacks and 5.69 for Whites, F (1,814) = 3.82, p = .05); 3) AIDS-Coercive Attitudes: Blacks were more likely to endorse coercive policies (mean scores = 3.39 for Blacks and 2.37 for Whites, F (1,884) = 16.85, p < .001); 4) AIDS-Pragmatism: Blacks were more likely to endorse nonmoralistic pragmatic policies (mean scores = 6.65 for Blacks and 5.99 for Whites, F (1,885) = 3.79, p = .05); 5) Attendance at religious services: on the 9-point scale of religious attendance, mean scores were 5.26 (indicating attendance 2-3 times per month) for Blacks and 4.2 (attendance once per month) for Whites (F [1,874] = 13.78, p < .001).

Correlation and Regression Analyses[4]

Correlations. Endorsement of coercive policies (high AIDS-Coercive scores) was associated with high Casual Contact Beliefs scores (r = .45, p < .001), high ATG scores (r = .42, p < .001), lower levels of formal education (r = -.20, p < .001), and greater age (r = .20, p < .001). Endorsement of pragmatic policies (high AIDS-Pragmatism scores) was associated with lower ATG scores (r = -.39, p < .001), lower age (r = -.21, p < .001), and (slightly) with high Casual Contact Beliefs scores (r = -.14, p < .001). AIDS-Pragmatism attitudes were not significantly correlated with Intimate Contact Knowledge scores (r = .05). AIDS-Pragmatism scores were more strongly correlated with AIDS Coercive scores than in the focus group sample (r = -.25, p < .001).

Negative attitudes toward gay men (high ATG scores) were associated with lower levels of education (r = -.25, p < .001) and with overestimation of the risks of HIV transmission associated with casual contact (r = .24, p < .001). High Casual Contact Beliefs scores were associated with lower levels of formal education (r = -.21, p < .001). In contrast to the focus group sample, Intimate Contact Knowledge scores were not significantly correlated with Casual Contact Beliefs scores (r = -.02).

Regression Analyses. As with the focus groups, separate regression analyses were conducted for AIDS-Coercion and AIDS-pragmatism attitude scores (see Table 4). Once again, different variables emerged as significant (p < .05) predictors of the two dimensions of attitudes. The bulk of variance in AIDS-Coercion scores was explained by Casual Contact Beliefs scores (9.9%) and ATG scores (6.5%); supporting coercive policies was associated with overestimating the risks posed by casual contact and expressing hostility toward gay men. Additional variance was accounted for by age (3.6%); older respondents were more likely than others to express support for coercive policies. In contrast to the focus group sample, the best predictor of AIDS-Pragmatism scores in the national sample was attitudes toward gay men (8.1% of the variance).

When the regression equations were recalculated separately for Blacks and Whites, different variables emerged as predictors of AIDS-related attitudes (see Table 5). For both groups, overestimation of the risks of casual contact was the principal predictor of support for coercive policies, accounting for 10.2% of the variance in Whites' AIDS Coercive scores and 9.8% in those of Blacks. For Blacks, the other significant predictor was religious attendance, which accounted for 5% of the variance. Blacks who attended church frequently were more likely than nonreligious Blacks to support coercive policies. For Whites, in contrast, the secondary predictor was anti-gay attitudes, which accounted for 8.2% of the variance. For both Blacks and Whites, additional variance was explained by the demographic variables of age (3.5% for Blacks and 3.6% for Whites), sex (2.8% for Blacks and 1.4% for Whites), and educational level (2.3% and 2.9%). Males who were older and less educated were more likely than others to support coercive policies.

Differences were also observed in the regression analyses for AIDS-Pragmatism attitudes. Attitudes toward gay men were most predictive of Whites' attitudes, accounting for 9% of the variance. Additional variance was explained by age (4.8%) and sex (2.4%); scores on Casual Contact Beliefs accounted for less than 1% of the variance. In other words, Whites were more likely to endorse pragmatic policies if they expressed positive attitudes toward gay men, and if they were younger and female. None of the demographic, attitudinal, or belief variables predicted Blacks' AIDS-Pragmatism attitudes.

A possible explanation for the racial differences is that, because 90% of the Black respondents belonged to a conservative religious denomination (versus 78% of Whites), race might actually have served as a proxy variable for religious conservatism in our sample. To test this hypothesis, additional regression analyses were conducted for the entire sample for both AIDS attitude scales. Based on earlier research by Herek (1987a), religious affiliation was coded as either orthodox/conservative (Baptist, "Born again" Christian, Catholic, Christian fundamentalist, Conservative Jewish, Mormon, Lutheran, Methodist, Orthodox Jewish, Pentecostal, Seventh Day Adventist) or liberal (Congregational, Episcopal, Presybterian, Reform Jewish, Unitarian, agnostic, atheist, and persons specifying "no religion"). For AIDS-Pragmatism scores, race accounted for 0.3% of the variance whereas religious denomination accounted for 1.3%. For AIDS-Coercion attitudes, race accounted for 1.6% of the variance and religious denomination for 0.5%. Thus, religious background appears to explain much of the racial difference in AIDS-Pragmatism attitudes, but not in AIDS-Coercion attitudes.

As Table 5 suggests, after Casual Contact Beliefs, Blacks' scores for AIDS-Coercion were predicted by frequency of attendance at religious services whereas Whites' attitudes were predicted by attitudes toward gay men. In order to assess whether the racial differences in regression coefficients were statistically significant, we constructed a series of regression equations with two steps. On the first step, a dichotomized race variable was entered (Whites coded as -1, Blacks as +1) along with one of the independent variables listed in Table 5 (e.g., ATG scores); on the second step, the multiplicative interaction term (e.g., Race X ATG) was entered. If the multiplicative term increased the proportion of variance explained, then it can be concluded that the regression coefficient for the continuous variable (e.g., ATG) differs significantly between the two groups described by the categorical variable, i.e., Blacks and Whites (see Pedhazur, 1982, Chapter 12). This procedure was followed with AIDS-Pragmatism scores and AIDS-Coercion scores as dependent variables, and with the independent variables of ATG, Casual Contact Beliefs, age, religious attendance, education, and religious denomination. For AIDS-Pragmatism scores, the only variables to show significant (p < .05) Black-White differences were ATG scores and Casual Contact Beliefs. For AIDS-Coercion attitudes, significant differences were observed only on ATG scores. As is evident in Table 5, both of these variables (especially ATG) exerted a stronger effect for White respondents' attitudes than for those of Blacks.[5]

Finally, because scores on the Casual Contact Beliefs scale were of such importance in explaining the variance in AIDS-Coercion attitudes, they were subjected to a separate regression analysis. Included among the independent variables (along with sex, age, and education) were attitudes toward gay men, religiosity, and responses to three items assessing trust in governmental and health authorities. For both Whites and Blacks, the principal predictor of high scores on Casual Contact Beliefs was distrust of scientists and doctors who say that AIDS is not spread by casual contact (accounting for 12.5% for the variance in Whites' scores and 7.9% in those of Blacks). The secondary predictors differed for the two groups. For Blacks, high Casual Contact Beliefs scores were predicted also by believing that the government is not telling the whole story about AIDS (2% of variance). For Whites, this variable also explained some variance (1.3%), but of greater importance were the variables of anti-gay attitudes (2.6%), education (4.5%) and age (3.5%). Finally, when AIDS attitude scores were entered into the equation on the final step (after all of the previously mentioned variables), AIDS-Coercion attitudes accounted for an additional 6.5% of the variance in Blacks' scores, and 5.5% in those of Whites. AIDS-Pragmatism attitudes did not explain a significant amount of variance for either group. Although we have been discussing scores on the Casual Contact Beliefs items as a predictor of AIDS-related attitudes, the relationship probably is reciprocal, at least for AIDS-Coercion attitudes.

General Discussion

Results obtained with the national sample generally support and extend our tentative conclusions from the focus groups: The two-factor structure of AIDS-related attitudes was replicated; the factors were not highly correlated; the factors appear to have somewhat different antecedents, which appear to differ between Blacks and Whites. Each of these conclusions is discussed briefly here.

The Psychological Dimensions of AIDS-Related Attitudes

The two dimensions of AIDS-related attitudes appear to be best described by the labels of PRAGMATISM/MORALISM and COERCION/COMPASSION. The former dimension consists of items advocating government policies to reduce HIV transmission risks associated with sexual behaviors and needle sharing; the items share a pragmatic focus on changing specific riskful behaviors rather than a moralistic advocacy of eliminating broad categories of behavior (e.g., homosexual behavior, heterosexual activity outside of marriage, illegal drug user. The latter dimension combines perceptions of HIV-infected individuals as menacing with attributions of blame. It is summarized in the assertion that people with AIDS are paying the price for their own behavior and must not be allowed to infect the "innocent." The belief that AIDS is itself a punishment (from God, from Nature) can be used to justify punitive measures such as quarantine within a larger ideology that the world is just and people get what they deserve (e.g., Lerner, 1970). This view probably also helps to reassure some of those promoting it: They can feel safe from this frightening epidemic because they have not engaged in blameful behavior. One reason for the disproportionate level of public attention in the United States to the small number of transfusion-related cases of HIV transmission may well be that such cases threaten the sense of safety and vulnerability afforded by the "just world" construction.

Similar conflicts between moralism and pragmatism and between coercion and compassion have been observed in public reactions to earlier health problems. Brandt (1987), for example, detailed the historical conflict between moralistic approaches and secular rationalism in government response to venereal diseases in the United States. Advocates of a secular rationalist approach typically recognized the inevitability of sexual behavior outside of marriage; they sought to reduce the incidence of venereal disease through distribution of prophylactics and, when effective antibiotics became available, through nonjudgmental treatment of infected individuals. Moralists, in contrast, advocated abstinence and appealed both to moral values and fear of disease to encourage it; they considered venereal disease to be symptomatic of deeper social and moral disorder. The parallel between these positions and the two extremes of the PRAGMATISM/MORALISM dimension described here for AIDS attitudes are striking.

Brandt (1987) also described a second historical conflict in societal responses to venereal disease, which may be related to the COERCION/COMPASSION dimension described here. He framed this conflict as pitting opposing views of responsibility: Is venereal disease (or AIDS) "merely the result of an individual's willful exposure, or should external, environmental, and social factors that might contribute to a tendency to exposure be considered?" (Brandt, 1987, p. 169; see also Brandt, 1988). Describing public discourse about herpes, Brandt (1987) observed, "The moral judgment is explicit; these diseases are received only through choice--a willful choice of questionable morals and mores. Moreover, the implication that . . . victims have got what they ultimately deserve runs beneath the surface of all these discussions" (p. 181). Behavior is assumed to be entirely voluntary and, once informed about risks, individuals are expected to modify their behaviors. "The assumption that an individual's behavior is free from external forces--that life-style is strictly voluntary--is explicit" (Brandt, 1987, p. 202).

Such individualistic views are not restricted to popular and political discourse; they also pervade social scientific research on health behaviors. The widely used Health Belief Model, for example, is based on individualistic and agentic assumptions about human nature; it focuses minimal attention on environmental and cultural factors that interfere with health-promoting behaviors or displace personal health promotion as a primary goal of individuals (e.g., Janz & Becker, 1984). Some AIDS-riskful behaviors, however, may not permit a "rational" decision-making process because their persistence reflects physiological addiction (e.g., the use of IV needles for drugs), non-health-related needs (e.g., needs for social approval or bonding, needs to express core aspects of one's identity), or environmental factors (e.g., distrust of government officials or experts based on a previous history of discrimination or repression). Failure to recognize the obstacles to individual behavior change created by such forces can cause observers (scientists included) to respond to riskful behavior with puzzlement and, eventually, hostility and coercion.

A Tentative Typology

The relatively low correlations between the PRAGMATISM/ MORALISM and COERCION/COMPASSION factors (r = -.13 for the focus group sample and -.25 for the national sample, based on simple additive scoring of the items) suggest the presence of four distinct response patterns (see Figure 1). These patterns correspond to the positions of some major constituencies and interest groups involved with the AIDS epidemic. First, a Compassionate Secularism pattern characterizes the general stance of the American public health community and of the lesbian and gay male community: endorsement of such nonmoralistic pragmatic policies as distribution of condoms and sterile needles, as well as opposition to coercive measures such as quarantine (e.g., Koop, 1986). In the national sample, 54% of Whites and 45% of Blacks manifested this pattern (i.e., they agreed with at least three AIDS-Pragmatism items and disagreed with at least three AIDS-Coercion items). Second, a pattern of Compassionate Moralism (operationally defined as disagreement with three or more items on each scale) is reflected in the official pronouncements of the National Conference of Catholic Bishops: Compassion is urged for people with AIDS, but education about condoms is rejected on moral grounds (e.g., Lattin, 1989). In our national sample, 20% of Whites and 21% of Blacks displayed this response pattern. Third, Punitive Moralism, endorsement of coercive measures and rejection of nonmoralistic pragmatic policies, is perhaps best exemplified in the United States by spokespersons of conservative political and religious groups, including the Religious Right (e.g., Buchanan, 1987; Cohen, 1987). In our sample, only 7% of Whites and 5% of Blacks displayed this response pattern.

Finally, 4% of Whites and 20% of Blacks agreed with at least three items on each AIDS attitude scale, a pattern we provisionally label Indiscriminate Action. This fourth pattern actually may combine several different perceptions of AIDS. It may reflect an acquiescent response set for at least some respondents; this interpretation will be tested in later research by formulating additional items that will be reversescored. It also may reflect a "do something" mentality, a willingness to endorse any AIDS-related policy that promises action, regardless of its likely costs, consequences, or effectiveness (Schneider, 1987). Such a mentality resembles hypervigilance, a coping pattern that results when decision-makers experience intense stress due to their perception that a) severe losses are imminent if current practices are not changed; b) losses also are imminent if current practices are changed; c) a satisfactory solution is possible; but d) insufficient time is available to search carefully for a solution. Janis (1989) summarized this pattern as "Try anything that looks promising to get the hell out of this agonizing dilemma as fast as you can. Never mind any other consequences" (p. 80). The Indiscriminate Action pattern also may reflect considerable ambivalence concerning AIDS: views of people with AIDS as both dangerous and deserving of compassion, views of societal responses to AIDS requiring containment as well as pragmatic education and prevention.

Antecedents of Attitudes

Different variables were observed to predict AIDS-Coercion and AIDS-Pragmatism scores. In the national sample, Whites' scores on AIDS Coercion were predicted principally by Casual Contact Beliefs and attitudes toward gay men. Whites were more likely to endorse punitive policies to the extent that they overestimated the risks of casual contact and expressed hostility toward gay men. Their scores on AIDS Pragmatism were shaped principally by their ATG scores: They were more likely to reject nonmoralistic pragmatic policies if they expressed anti-gay hostility. Blacks were similar to Whites in that Casual Contact Beliefs scores were an important predictor of AIDS-Coercion attitudes. They differed from Whites in that the secondary predictor of those attitudes was religious attendance rather than attitudes toward gay men. No variables were observed to predict AIDS-Pragmatism scores among Blacks.

We interpret these patterns to mean that Whites and Blacks in the United States generally perceive the AIDS epidemic in different ways. For Whites, AIDS is a disease of the "Other," strongly identified with gay men and homosexuality. Blacks, in contrast, may perceive AIDS as a problem of the African-American community. Because of historical patterns of racism in the United States, many Blacks probably distrust the government's role in the AIDS epidemic and their responses to the epidemic instead are shaped by community institutions such as churches (Dalton, 1989). Even though Black respondents' attitudes toward gay men were more negative than those of Whites, these attitudes may be relatively unimportant to AIDS attitudes because Blacks may not perceive gay people to be members of the Black community. Indeed, we suspect that many of the Black respondents (and probably many White respondents as well) translated "gay men" to mean "White gay men" (e.g., Icard, 1986).

The data on which these conclusions are based are limited in some respects. Not only was the number of Black respondents small (n = 81), the response rate for the national survey also was rather low (47%), in part because resources were lacking for follow-up calls to persuade "refusers" to agree to be interviewed. Another shortcoming is that response options for several scales included in the survey were limited to agree, disagree, or in the middle; a 5-item or 7-item Likerttype response scale would have permitted greater response variation which, in turn, would have increased our confidence in the factor analyses and multiple regressions.

Despite these limitations, the work described in this paper offers promising hypotheses for future research. The distinction between COERCION/COMPASSION and PRAGMATISM/MORALISM policy attitudes (which we now have observed in other samples in addition to those described here) offers a useful framework for conceptualizing AIDS-related attitudes. The finding that AIDS-related attitudes are influenced by different variables among Whites and Blacks suggests that public education programs should adopt strategies targeted to specific audiences. Reducing anti-gay prejudice appears to be a necessary prerequisite for influencing AIDS attitudes among Whites and, perhaps, for effectively communicating credible information about the lack of risk associated with casual contact. The primary and powerful psychological linkage between AIDS and gay men is likely to continue in the minds of Americans, even as increasing numbers of AIDS cases are manifested among heterosexual IV-drug users, their sexual partners and partners' partners, and infants born to infected women. (For further discussion on the levels of stigma and symbolism associated with AIDS and homosexuality, see Herek, 1990; Herek & Glunt, 1988; for discussions of anti-gay prejudice, see Herek, 1984, 1986, 1987a, 1987b, 1988). Among Blacks, AIDS-information programs must overcome deeply ingrained suspicions of government and of White-identified scientific experts (Dalton, 1989).

One approach might be to encourage educational programs through Black community institutions, such as churches, which generally have been slow to respond to the epidemic (Lambert, 1989; Shilts, 1989). In this process, AIDS-educators should be careful to distinguish knowledge from beliefs. Many Blacks (and undoubtedly many Whites as well) know the "official story" about HIV transmission but do not believe it. To be effective, educators must instill trust as well as impart information. Because of sampling limitations in the present study, we offer this interpretation as a hypothesis for future testing rather than as a conclusion firmly supported by data. Understanding the social and psychological influences on AIDS-related attitudes among African-Americans is especially important, given their overrepresentation among reported cases of AIDS in the United States.[6]

Understanding public attitudes concerning AIDS requires sorting through the various and competing motives for those attitudes: Fear and compassion; prejudice and tolerance; concerns for public health and desire to protect civil liberties; personal values concerning stigmatized conduct and concern for saving lives. The present study offers preliminary direction in making sense of these motives.

[1] The items included in the scale are: eating food in a restaurant; using a public telephone; using a public toilet; being in a place where persons with AIDS gather; standing near a person with AIDS who is coughing or sneezing; sharing a drinking glass; being bitten by a mosquito; shaking hands with a person with AIDS.

[2] Calculation of response rate was based on the formula:

R = C/T, where R = response rate.

C = number of completed interviews (1,078).

T = total number of eligible numbers sampled (2,302) (this is equal to the total number of telephone numbers sampled, minus non-working numbers, minus nonresidential numbers, minus households without any eligible respondents, minus numbers with no answer after at least 20 calls).

[3] The transmission routes were: being bitten by a mosquito; sharing a drinking glass with someone you don't know; having sexual intercourse without using a condom; having sexual intercourse while using a condom; receiving a blood transfusion; giving blood to a blood bank; using a public toilet; using a public telephone; eating food in a restaurant; standing near a person with AIDS who is coughing or sneezing; shaking hands with a person with AIDS; working in the same room as a person with AIDS. The response options were YES (i.e., AIDS could be transmitted through this router, NO (AIDS could not be transmitted), and MAYBE.

[4] Because of the sample size, relatively small correlation coefficients achieved statistical significance. Only the larger, substantively significant coefficients (> .20) are highlighted here. Similarly, independent variables are highlighted in discussing the regression analyses only when they explained at least 1% of the variance in the dependent variable.

[5] Similar comparisons between regression coefficients for the independent variables were made between White males and females and between Black males and females. No significant sex differences were observed.

[6] The same is true of Hispanic-Americans, a minority not sufficiently represented in our samples to permit separate analysis.

Table 1

Factor Loadings for AIDS-Related Policy Attitude Items (Focus Group

Sample)

PART I

COERCION/

COMPASSION

Item (16.6%)

The government should give away condoms ("rubbers") to stop the spread of AIDS --

The federal government should pay for educational programs to teach people how to have "safer sex." --

The federal government should spend more money for research on AIDS. --

Our country needs civil rights laws to protect people with AIDS from discrimination. --

The government should fight AIDS among drug users by giving clean needles to anyone who wants them. --

Parents should not have to send their children to a school where another child with AIDS is enrolled. .7122

People with AIDS should be legally guarantined to protect the public health. .6963

More effort should go to testing people for the AIDS virus than should go to public education about AIDS. .5650

The names of people with AIDS should be published in newspapers so that others can avoid them. .5545

All people at high risk for AIDS should be required to take the test for AIDS-antibodies. .4996

People who want to quarantine persons with AIDS are just showing their own bigotry. --

I would accept a group home in my neighborhood where people with AIDS could live and get good care. --

Scientists who say that AIDS isn't spread by casual contact don't really know as much as they claim. .5263

PART II

PRAGMATISM/MORALISM

Item (22.5%)

The government should give away condoms ("rubbers") to stop the spread of AIDS .7557

The federal government should pay for educational programs to teach people how to have "safer sex." .7431

The federal government should spend more money for research on AIDS. .6430

Our country needs civil rights laws to protect people with AIDS from discrimination. .5945

The government should fight AIDS among drug users by giving clean needles to anyone who wants them. .5629

Parents should not have to send their children to a school where another child with AIDS is enrolled. --

People with AIDS should be legally guarantined to protect the public health. --

More effort should go to testing people for the AIDS virus than should go to public education about AIDS. --

The names of people with AIDS should be published in newspapers so that others can avoid them. --

All people at high risk for AIDS should be required to take the test for AIDS-antibodies. --

People who want to quarantine persons with AIDS are just showing their own bigotry. --

I would accept a group home in my neighborhood where people with AIDS could live and get good care. .4534

Scientists who say that AIDS isn't spread by casual contact don't really know as much as they claim. --

Note: Loadings less than .45 are omitted. Loadings are for a principal components analysis with varimax rotation (with oblique rotation, inter-factor correlation = -.11). Responses are on a 5-point Likert-type scale, ranging from strongly disagree to strongly agree (n = 143). The final three items were not used in the AIDS attitude scales.

Table 2

Regression Coefficients: Focus Group Sample

PART I

Unstandardized Standardized

Variable (b) (beta)

AIDS-COERCION ATTITUDES

Sex n.s. n.s.

Race n.s. n.s.

Education -0.4317 -.2380

ATG 0.2447 .3282

Casual contact beliefs 0.2290 .3359

Religious attendance -.0604 n.s.

Intimate contact knowledge n.s. n.s.

PART II

Variable T

AIDS-COERCION ATTITUDES

Sex n.s.

Race n.s.

Education -2.957 (p < .01)

ATG 3.680 (p < .001)

Casual contact beliefs 3.968 (p < .001)

Religious attendance n.s.

Intimate contact knowledge n.s.

For Equation: R[sup 2] = .3819 F(7,104) = 9.179 (p < .001)

PART III

AIDS-PRAGMATISM ATTITUDES Unstandardized Standardized

Sex n.s. n.s.

Race n.s. n.s.

Education -0.1493 n.s.

ATG -0.0574 n.s.

Casual contact beliefs -0.0800 n.s.

Religious attendance -1.2565 -.3482

Intimate contact knowledge n.s. n.s.

PART IV

Sex n.s.

Race n.s.

Education n.s.

ATG n.s.

Casual contact beliefs n.s.

Religious attendance -3.387 (p < .001)

Intimate contact knowledge n.s.

For Equation: R[sup 2] = .1527 (F(7,104) = 2.667 (p <.05)

Table 3

AIDS-Related Attitudes: Factor Loadings and Response Distribution

(National Sample)

PART I

Agree

PRAGMATISM/MORALISM (11.6% of variance)

The government should give away condoms to stop the spread of AIDS (.7289) 47.6

The government should pay for programs to teach people how to have "safer sex." (.6695) 68.7

The government should fight AIDS among drug users by giving clean needles to anyone who wants them. (.5828) 31.8

The federal government should spend more money for research on AIDS even if it means raising taxes. (.5176) 69.6

Our country needs laws to protect people with AIDS from discrimination. (.5016) 70.5

COERCION/COMPASSION (19% of variance)

People with AIDS should be legally separated to protect the public health. (.7439) 19.5

The names of people with AIDS should be published in newspapers. (.6453) 9.6

People with AIDS are getting what they deserve. (.6139) 10.9

People with AIDS are a serious risk to the rest of society. (.5713) 51.6

People with AIDS have only themselves to blame. (.5134) 19.1

PART II

Disagree

PRAGMATISM/MORALISM (11.6% of variance)

The government should give away condoms to stop the spread of AIDS (.7289) 47.9

The government should pay for programs to teach people how to have "safer sex." (.6695) 25.5

The government should fight AIDS among drug users by giving clean needles to anyone who wants them. (.5828) 61.8

The federal government should spend more money for research on AIDS even if it means raising taxes. (.5176) 25.5

Our country needs laws to protect people with AIDS from discrimination. (.5016) 21.3

COERCION/COMPASSION (19% of variance)

People with AIDS should be legally separated to protect the public health. (.7439) 72.5

The names of people with AIDS should be published in newspapers. (.6453) 86.7

People with AIDS are getting what they deserve. (.6139) 80.0

People with AIDS are a serious risk to the rest of society. (.5713) 35.0

People with AIDS have only themselves to blame. (.5134) 67.9

PART III

Not Sure

PRAGMATISM/MORALISM (11.6% of variance)

The government should give away condoms to stop the spread of AIDS (.7289) 4.2

The government should pay for programs to teach people how to have "safer sex." (.6695) 5.7

The government should fight AIDS among drug users by giving clean needles to anyone who wants them. (.5828) 5.6

The federal government should spend more money for research on AIDS even if it means raising taxes. (.5176) 5.0

Our country needs laws to protect people with AIDS from discrimination. (.5016) 7.5

COERCION/COMPASSION (19% of variance)

People with AIDS should be legally separated to protect the public health. (.7439) 7.8

The names of people with AIDS should be published in newspapers. (.6453) 3.6

People with AIDS are getting what they deserve. (.6139) 8.5

People with AIDS are a serious risk to the rest of society. (.5713) 13.0

People with AIDS have only themselves to blame. (.5134) 12.5

Note: Numbers in parentheses are factor pattern loadings for a principal components analysis with oblique rotation (n = 925; interfactor correlation = -.17); each item loaded highly (> .44) only on the factor under which the item is listed in the table. Five additional items did not load highly on either factor and are omitted from the table. Proportion of responses in each category are weighted by sex, race, and age, based on the U.S. Census Bureau's Current Population Survey for March, 1988. Margin of error due to sampling: +/-3.

Table 4

Regression Coefficients: National Sample (Whites and Blacks combined)

PART I

Unstandardized Standardized

Variable (b) (beta)

COERCION/COMPASSION ATTITUDES

Casual Contact Beliefs 0.2556 .3377

ATG 0.1969 .2915

Age 0.0430 .2705

Religious attendance 0.0450 --

Sex -0.8993 --

Race -0.0805 --

Education -0.1454 --

Religious denomination -0.0040 --

Sex X race 0.6619 --

Education X race 0.1025 --

Age X race -0.0300 --

PART II

Variable T

COERCION/COMPASSION ATTITUDES

Casual Contact Beliefs 11.118[c]

ATG 8.991[c]

Age 2.374[a]

Religious attendance n.s.

Sex n.s.

Race n.s.

Education n.s.

Religious denomination n.s.

Sex X race n.s.

Education X race n.s.

Age X race n.s.

For Equation: R[sup 2] (adj) = .3067 F(11,854) = 35.782[c]

PART III

Unstandardized Standardized

(b) (beta)

PRAGMATISM/MORALISM ATTITUDES

Casual Contact Beliefs -0.0498 --

ATG -0.2507 -.3266

Age -0.0030 --

Religious attendance -0.0368 --

Sex -0.6519 --

Race -1.5459 --

Education -0.2546 --

Religious denomination -0.3040 --

Sex X race 1.1641 --

Education x race 0.2299 --

Age x race -0.0262 --

PART IV

n.s.

Casual Contact Beliefs -9.305[c]

ATG n.s.

Age n.s.

Religious attendance n.s.

Sex n.s.

Race n.s.

Education n.s.

Religious denomination n.s.

Sex X race n.s.

Education x race n.s.

Age x race n.s.

For Equation: R[sup 2] (adj) = .1872 F(11,854) = 19.105[c]

Note: Unstandardized regression coefficients are reproduced for all variables to permit comparison across equations; standardized regression coefficients and T-values are reproduced only when statistically significant. Coefficients reflect final step of the regression equation with all variables included. Mean scores were substituted for missing data in independent variables; listwise deletion of missing data yields similar results.

Table 5

Regression Coefficients Blacks' and Whites' AIDS Attitudes, National Sample

PART I

Whites (n = 784)