Version classiqueVersion mobile

Sexual Behaviour and Risks of HIV Infection

Michel Hubert

I. Understanding Sexual Behaviour: Some Perspectives in the Context of HIV Infection

HIV Risk Perception and Determinants of Sexual Behaviour

Michaël Pollak et Jean-Paul Moatti

Texte intégral

This paper is based on research which has greatly benefited from cooperation with Dr W. Dab (Observatoire Régional de la Santé d'Ile de France), N. Bajos and C. Serrand (Comité Français d’Education pour la Santé).

1Summarising current knowledge about risk perception and modifications of sexual behaviour related to HIV infection is a rather difficult task. On the one hand, there has been a huge amount of social science literature on risk perception dealing with a variety of fields (attitudes and behaviours in the face of uncertain events, technological risks, individual risks for health, risk communications during emergencies and disasters, etc.). On the other hand, research experience (including ours) on the determinants of sexual behaviour in the context of the AIDS epidemic is still scarce. The existing data come mainly from studies of homosexual/bisexual men, the results of which simply cannot be extrapolated to other groups; for moreover, studies are very often epidemiologic follow-ups of cohorts concerned mainly with the magnitude of behavioural change without explanatory attempts. A recent review of the literature on AIDS and behavioural change to reduce risk concluded that "the immediate question becomes why most individuals successfully change their behaviours while some others do not" and that "distressingly little attention has been paid to this central issue" (Becker & Joseph, 1988).

2In the first part of this contribution to the workshop, we will present briefly some conclusions that can be derived from the general literature on risk perception and its relation to individual behaviour and behaviour change toward risks for health and safety and discuss their limits in the light of the issues raised by the AIDS epidemic. In the second part, we will try to discuss some basic problems of research on the determinants of sexual behaviour and reactions to the risks of sexual transmission of HIV. We will illustrate these points chiefly with material from our own surveys in France (see Annex I).

3It is beyond the scope of this article to decide if the specificities of sexual behaviour and HIV infection imply only an adaptation and enrichment of existing explanatory models of behavioural change or if they create a necessity for new models and perhaps new theoretical paradigms for analysing human attitudes and behaviour in the field of sexuality. However, our still very subjective preference is for the latter hypothesis.

I. The limits of traditional models about determinants of behaviour in the case of sexual HIV transmission

4Consensus will be easily reached among social scientists about some general conclusions drawn from the literature on behavioural risk modification, especially through health information and education, and that have already been confirmed by studies on AIDS.

51. There is no direct relationship between an individual's level of knowledge and attitudes toward a disease and behaviour. Information alone is therefore insufficient to promote meaningful change in risk behaviour, especially when that behaviour is immediatly reinforcing and well established, and its negative consequences are temporally distant or uncertain. Illustrations of this point are numerous in studies about different health-related problems such as smoking, drug and alcohol abuse patterns, and obesity. In the field of sexual behaviour, one can refer to research on the impact of programs to prevent teenage pregnancy showing that whilst information on birth control significantly increases knowledge about contraception, the likelihood of pregnancy and sexual activity are not consequently affected (Lance, 1975).

6Because sex is a powerful motive and because sexual practices are maintained by past experiences, immediate gratification, reinforcment by fantasies and often interpersonal influence or even coercion, it can be expected that sexual activities are especially difficult to change through the provision of information alone. One can easily agree with Nelkin (Nelkin, 1987) that the common examples of direct behavioural responses to information are "all in areas in which alternative choices are available so that changes in behaviour require no significant changes in life-style". All studies among homosexual/bisexual men have confirmed that individual knowledge about AIDS risk is not statistically related to reported frequency of high-risk sexual practices for HIV infection and observed behavioural change to reduce exposure (for a review, Kelly et al., 1988).

Table 1a: Beliefs of the general public about transmission of AIDS (Paris region, N=900, December 1987).

% believing in the mode of transmission

Sexual intercourse


IV drug injection


Blood transfusion


Using a carrier's razor


Blood donation


At the dentist


By a carrier's saliva


By a mosquito bite


In public lavatories


By kissing a carrier


By drinking from a carrier's glass


Hospitalization close to AIDS patients


At the swimming pool


Table 1b: Opinions of the general public about prevention of AIDS (Paris region, N=900, December 1987).

% who "agree" or "strongly agree"

Support sexual education in schools


Support advertising for condoms on TV


Support legal action against HIV carriers who consciously transmit the disease


Trust HIV carriers to inform partners


HIV mandatory screening for the whole population


Isolation of AIDS patients in hospitals


Withdrawal of AIDS children from school


Quarantine for AIDS patients


Quarantine for HIV carriers


7What may sometimes be confusing for public health decision-makers is that there is indeed a strong relationship between knowledge about AIDS and social perception of the disease, including opinions on its prevention. As shown in Tables 1 and 2 from our survey of the Paris-area population, misbeliefs about transmission through casual contact and by donating blood are related to individuals' willingness to support policies which carry a great danger of stigmatization for AIDS victims and HIV carriers, but there is still no relation between beliefs on transmission and reported "safe" behaviours such as use of condoms. Similarly, according to our sample of heterosexuals with multiple partners, the use of condoms is not significantly lower among individuals who think that washing or urinating after sexual intercourse or practicising coitus interruptus may be effective means of protection against sexual transmission of HIV.

Table 2: Cross-rate tabulations between beliefs of modes of transmission of HIV and opinions on prevention

% who agree with quarantining AIDS patients

% who agree with mandatory screening for the general population

Believe in transmission



by donating blood

Do not believe



Believe in transmission



at the dentist

Do not believe



Believe in transmission



in public lavatories

Do not believe



Believe in transmission



by drinking from a carrier's


Do not believe



Believe in transmission



by being hospitalized close

to AIDS patients

Do not believe



Believe in transmission



at the swimming pool

Do not believe



* p < 0.001
** p < 0.0001

82. Research on behavioural change has repeatedly shown that trust and credibility in the source of information and social reinforcement by peers within the community are necessary conditions for effective intervention. Early findings of experimental social psychology on "small group decision" have emphasized the importance of adaptation to ambiguity or change in the environment through interpersonal communication and referred to mechanisms such as group polarization and group pressure, social confirmation, norm formation and minority innovation (Bandura, 1969; Sherif, 1935). In one of the largest American cohort studies of homosexual/bisexual men, supportive peer norms appeared to be the only factor related to longitudinal behavioural change to reduce the risk of sexual HIV transmission (Emmons et al., 1986).

9However, when it comes to explanatory models for determinants of behaviour, the main (implicit or explicit) reference of most AIDS studies remains the Health Belief Model (HBM) which was established in the context of assessment of health education programmes for other diseases (with a special input from the internationally organized 10-year research project in North Karelia, Finland, on the prevention of coronary heart disease) (Puska et al., 1985; Jette et al., 1981; Janz & Becker, 1984). In its current formulation, the HBM assumes that the following factors mostly determine the likelihood that an individual will take a given preventive or curative action with regard to a particular health problems:

  • the perceived individual susceptibility to the disease;

  • the perceived seriousness of the disease;

  • the perceived benefits of the health action;

  • the perceived barriers to the health action;

  • certain cues for action;

  • modifying demographic and psycho-sociological factors (including the so-called "general health motivation").

10The combination of high perceived susceptibility to the illness and great perceived benefits of action is usually a good predictor of adoption of behavioural change; it is also often assumed (without experimental evidence) that components of the model combine in a multiplicative manner to influence behaviour.

11It does not seem unfair to consider the HBM as an attempt to build an operational application of classical social psychological learning theories to health education, the assumption being that health beliefs are modifiable and such modifications of beliefs will induce risk-reducing behavioural change.

12It cannot be denied that valuable practical recommendations for health education have been derived from the HBM (or at least from empirical research referring to its conceptual framework): for instance, effective health/behaviour-change educating messages must include explicit information indicating that the severity of the potential illness is great, the individual receiving the message is susceptible to the illness, a change in behaviour can reduce the likelihood of illness, and the relative benefits of behaviour change are greater than the costs (including psychological costs). The messages' effectiveness is also increased when they simultaneously provide information on the specific behavioural change needed to reduce risk, offer a cognitive rationale for the reasons why theses changes produce practical reduction of risks, and provide positive encouragement for making health-related changes (Siegel et al., 1986). Whilst the importance of exhaustive, accurate information is commonly accepted, discussion is still open on the most effective content of messages, such as the role of fear in risk awareness (Job, 1988). Whilst it is clear that prevention campaigns based only on fear are likely to prove ineffective and that individuals overwhelmed by fear are more likely to feel that risk exposure is unavoidable, it is often argued that messages producing moderate levels of fear will facilitate behavioural change to the extent that they are completed by a positive, reassuring description of the consequences of change.

A. General limits of the Health Belief Model (HBM)

13Still, when it comes to the explanatory power of determinants of behaviour from the HBM, many criticisms may be raised. Once the HBM is no longer used for practical purposes (designing and evaluating health education programs), its explanatory power easily looks like a tautology.

14In particular, it is obvious that the association between health beliefs and behaviour is bidirectional, and that health beliefs are as much a consequence of behaviour as its cause. So, it is always possible to find an a posteriori relationship between observed behavioural change (or absence of change) and a general model of perceived risk/perceived benefit. For example, in our sample of heterosexuals with multiple partners, we found significant relationships between individual perception of being at risk for HIV, individual fear of STDs, beliefs in the effectiveness of condoms to protect againts STD's and HIV transmission, and reported use of condoms (Table 3).

15However, it is impossible to determine whether these perceptions are real determinants of condom use or only a posteriori rationalizations of established use or reduction of use. Similarly, some studies of homosexual men showed that imagining the physical deterioration associated with AIDS "predicts" a significant reduction in the number of sexual partners and highrisk sexual practices. Here too, one might ask to what extent it is theoretically legitimate to subsume the empirical data referring to this particular capacity of imagination into the general concept of risk perception.

16Beyond the traditional technical problem of the adequation of the psychometric scales used to quantify risk perception, the more basic problem of what exactly is measured under the item of risk perception and its determinants remains. Most studies of AIDS, when they are not limited to labelling some expressed beliefs under the "risk perception" item but genuinely try to build adequate scales of risk perception, have failed to show a relation "in any straightforward fashion to either cross-sectional or longitudinal findings" (Becker & Joseph, 1988). A pure measure of riskperception of AIDS, independantly of the individuals' histories of sexual practices and their proximity (real or symbolic) to the disease seems, in any case, impossible to obtain. For example, in a sample (n = 400) of pregnant women attending their first prenatal visit in two Paris public hospitals, we found that the proportion of women classifying themselves as "at above-average risk" of being contaminated by HIV was the highest (15.5%) among women who "thought about having a screening test for HIV but did not have it" than among those who had recently been tested (9.5%) or had not been tested but had not previously thought about it (2.8%) (Moatti et al., 1989c).

Table 3: Risk perception and condom use among French heterosexuals with multiple partners (CFES-INSERM, N=1000, Sept. 1988)

% of condom users

Individual risk perception

of contracting AIDS:

No risk


Small risk

46.3 *

Average risk


High risk


Individual fear of STDs:

No fear


Average fear

53.3 **

High fear


Beliefs in effectiveness of condoms:

Always effective


Sometimes ineffective

45.1 **

Often ineffective


* p < 0.05
** p < 0.001

  • 1 In the literature on risk perception, such mental processes are often qualified as "biases" to expl (...)

17Moreover, the HBM seems to ignore the vast body of economic and psychological research on individual mechanisms of risk perception that has been carried out to test the validity of the axioms of expected utility theory. This research has identified a number of very general inference rules that individuals seem to use when evaluating risk for their own health. This research is primarily concerned with high uncertainty situations. These judgmental rules, called heuristics by Tversky and Kahneman (1974), are employed to reduce difficult mental tasks to simpler ones1. The "availability heuristic" (tendancy to judge an event as likely or frequent if instances of it are easy to imagine or recall) is of special relevance to risk perception; the effects of memorability and imaginability may pose a barrier to open consciousness of risk while lack of availability may lull people into complacency.

18Heuristics produce individual overconfidence in judgments based upon them, desire for certainty (and consequent denial of low-probability risk), and the well-known "it won’t happen to me" form of overconfidence (that is, the fact that people tend to consider themselves personnaly immune to many hazards whose societal risks they would readily acknowledge).

19Collective influences on individual mechanisms of risk perception have also been pointed out by research on comparative estimation of different risks for health, especially in the field of technological risks (Fischhoff et al., 1982). Three characteristics maximize risk aversion, namely unfamiliarity, catastrophic potential (simultaneous exposure in space and time of a high number of individuals) and lack of control (Lefaure & Moatti, 1983). This last factor does not refer to the traditional distinction between voluntary (and therefore more acceptable)/unvoluntary risk, which has been proved invalid (Slovic et al., 1980), but rather to the individual feeling that society or the community as a whole is able to control the risk and its detrimental consequences.

20Finally, at a societal macrolevel, risk perceptions are heavily influenced by political, cultural and social factors, such as membership in social class or involvement in a social milieu. Attitudes toward risk are highly embedded in the system of beliefs, values, and ideals that constitutes a culture. Thus, different cultures and social groups will emphasize certain risks and minimize others. Perceptions of risk are also closely connected to legitimating of moral principles (Douglas & Wildavsky, 1982). A judgment about risk can be a social comment, reflecting points of tension and value conflicts in a given society. Our own work on technological risks has suggested that expressed risk perceptions are sometimes the product of ideological and ethical precommitments rather than the product of the perception of the characteristics of an activity and its associated risks per se (Bastide et al., 1989). This hypothesis also seems adequate for some aspects of AIDS (Moatti et al., 1988).

  • 2 A fundamental epistemological difference is that expected uility theory, at least in economics, doe (...)

21The previous considerations lead to the most fundamental criticism of the Health Belief Model. In a way, this model is not different from the conventional axioms of expected utility theory, which are based on the assumption of rational individuals acting as utility (or risk/benefit, cost/benefit ratios) maximizers2.

22Being centred on the individual, the HBM neglects the dynamic social interactions that shape behaviour. Alternative research approaches may come from recent developments in "social identification theory" that have extended the basic model of the "social self" as a mix of internalized group perspectives. Turner (1982) has proposed the idea that self-concept is an organized cognitive structure but that its functionning is adaptive and situation specific. In other words, in a given situation, people may adopt one of any number of possible selves, activating some salient aspects of the personal and social identities. Some behavioural situations (notably competition or intergoup conflict, but why not also emotional/sexual relationships?) engage social identity and activate identification with a particular group.

23In any case, research on sexual behaviour and the risk of HIV infection clearly requires focusing on the specific social interactions that influence individual behaviour, which is largely absent from the HBM.

B. Specific limits in the case of HIV infection

24Many empirical data from AIDS studies emphasize the need to analyse individual behaviour as a dynamic product of various levels of social interaction. Some examples are:

  • the fact that studies on behavioural change in intravenous drug users show a lot more significant changes concerning risk of transmission through needle-sharing than by adopting "safer sex" practices (Becker & Joseph, 1988; Kelly et al., 1988);

  • the apparent distinction, in terms of high-risk practices, between, what is permitted within a primary relationship and what is allowed outside such a relationship in most studies of homosexual/bisexual men (Becker & Joseph, 1988; Poliak & Schiltz, 1987);

  • the expressed attitudes toward condoms among users and non-users. In our sample of heterosexuals with multiple partners, condom users and nonusers do not differ in their opinions about the direct consequences of condoms on sexual intercourse but rather on the social image of the condom and its potential impact on affective and sexual relationships (Table 4).

  • 3 CFES-INSERM, N=1000, September 1988.

Table 4: Opinions about condoms (French heterosexuals with multiple partners)3

% who agree with the following statements about condoms

Condom users

Condom non users

Are difficult to use




Reduce sexual pleasure




Induce sexual impotence




Limit one’s sexual pleasure




Limit partners' sexual pleasure




Kill romance in relationship




Create doubts about partner




Are shameful to buy




Are ordinary consumer goods




* p <0.05 ** p <0.001 *** p <0.0001

25More detailed arguments from our surveys in France are presented below.

II. Attitudes and behavioural changes in the male homosexual population

26Our first survey among homosexuals showed the contrast between a very high degree of information and knowledge on the one hand and the absence of sexual precautions on the other. If, in 1985, almost all French gay men understood correctly the nature of the disease and its ways of transmission, only 5% admitted to using condoms. This situation changed rapidly after 1986, when the introduction of the HIV antibody test and its massive voluntary use by homosexuals made visible the size of the epidemic and increased concernedness. Condom use became widespread and a significant number of gay men (13%) gave up anal intercourse. At the same time, it became obvious that macrosociological factors such as class and education have a major influence on changes in sexual behaviour.

27Behavioural risk adaptation also changes over time. In general, people start by reducing the number of situations of exposure (avoiding some places such as backrooms and bath-houses, reducing the number of partners, extending the length of steady relationships, practising sexual abstinence) before taking precautions that interrupt virus transmission (abandoning anal penetration, using condoms).

28After four years of observations, we have found the following adaptative models to risk of HIV transmission (Poliak, 1988).

291. Although still in the small minority, the proportion of gay men without sexual relations has increased from 5 to 8%.

302. From 1985 to 1987 we found a simultaneous reduction in the number of sexual partners and a marked increase in steady couple relationship (from 10 to 25%) as well as a significant decrease in the number of respondents with more than 20 sexual partners over a period of six months. What do these changes tell us about "gay promiscuity"? There is no general trend towards couple relationships - the proportion of respondents indicating a privileged partner relationship has remained stable over time (50%). The increase in "closed couples" indicates that people with such a privileged relationship have significantly reduced their "outside relationships". In addition, only the most promiscuous group has reduced the number of partners; the proportion of gay men with two to ten partners has remained unchanged (50%).

Table 5: Number of sexual partners during the six months before survey (French male homosexuals / bisexuals)

of partners















2 to 5





6 to 10





11 to 20





> 20





313. Precautions interrupting the virus transmission have increased rapidly: 13% of the sample gave up anal intercourse before 1986 and condom use has become widespread and increasingly regular (39% in 1988 used condoms regularly, 29% irregularly).

32Over the same period, voluntary testing has increased from 33% in 1986 to 59% in 1988 with a proportion of 19% HIV antibody positive results in 1988 (self-reported test results giving an estimate of seroprevalence in the gay group).

Figure 1: Behavioural changes among French homo/bisexuals

Figure 1: Behavioural changes among French homo/bisexuals

33More sophisticated statistical analysis allows us to relate these changes to relatively homogenous subgroups that we could isolate in our 1985 survey and observe since then:

  1. Homosexuals living in localities of less than 20,000 inhabitants, often hiding their sexual preference and whose homosexuality is not accepted by family and colleagues, feared social discrimination as much if not more than being contaminated. Desolidarisation with a "gay group and destiny" was widespread; the individual's ability to change behaviour was limited and the spread of condom use slow before 1987; willingness to support repressive measures was high (compulsory testing, quarantining). A feeling of being unable to protect themselves with adequate behaviour changes explains the demand for state intervention, which declined rapidly, even in this subgroup, as condom use became more widespread.

  2. Blue-collar gays felt hardly concerned by AIDS in 1985. Only after 1987 did they recognise the risk and start to change sexual behaviour.

  3. Lower-middle-class gays in big cities (service sector) showed the highest degree of denial. Although well informed about the disease, they often presented the "risk group" classification as an attempt to discriminate against them. Their biographical experience, including breaking up with their families because of their being gay, explains their ambivalent reactions in the early days of the epidemic. Only in 1986 did they start to change their behaviour.

  4. Upper-middle-class gays, particularly those with intellectual professions, were the first to adapt quickly to the risk of HIV transmission by giving up anal intercourse or by using condoms. They were also the first to know HIV carriers in their personal environments.

  5. Before 1986, scepticism prevailed among young homosexuals under 25. They identified AIDS with a disease of their parents' generation and resented the labelling of the "young" as a specifically exposed group. After 1986, differences appeared in this age group, with students following he upper-middle-class model and the rest the blue-collar model.

Figure 2: Correspondence analysis of answers to 1985 survey in French homosexual community

Figure 2: Correspondence analysis of answers to 1985 survey in French homosexual community

34These empirical observations suggest that one has to differentiate between various levels offactors that favour or hinder sexual behaviour change induced by risk of HIV infection (Table 6), viz.,

  • individual predispositions;

  • factors of the immediate environment (community-specific);

  • macrosocietal determinants.

Table 6: Factors most related to sexual behaviour change or hindering it in French homosexuals




Personal proximity

Widespread use of test creates visibility and symbolic proximity

Factors limiting further diffusion:
- low educational level
- social class

Social acceptance and self-esteem

Community belonging

Relations with medical authorities


Higher middle class fl

All middle classes fl

Limits on further spread

35We also can distinguish three different phases in the process of change.

36Phase 1 (before 1986): emergence of "pioneers" of new sexual behaviours among the ones who are the first to know personally people affected by the new disease. To this first condition for a change in consciousness one has to add factors reinforcing their predisposition and capacity to change, e. g., selfesteem and the feeling of being socially accepted; confidence in medical authorities and concernedness with health. Although their approach may look like the result of an individual rational "decision", it is in fact dictated by their middle-class status and high educational level.

37Phase 2 (1986-1988): The widespread use of the test makes the AIDS epidemic "visible". A feeling of identification with a collective risk and a new positive ethics of sexual precautions, beyond the boundaries of individuals personally acquainted with infected persons, emerges in particularly concerned groups, such as the gay community. The rapid spread of behaviour changes in all middle classes is favoured by this new sexual ethic and the identification with a gay destiny and community.

38Phase 3 (after 1988): After a process of rapid spread of sexual behaviour changes, one observes its limits. These are of a macrosocietal nature. Factors usually related to cumulative socio-economic handicaps and inequalities (low socio-economic status and fragility, such as unemployment, lack of confidence in the future, low educational level) play a major role in limiting adoption of adequate preventive behaviour.

III. Reactions to AIDS in the general population

39Whilst process analysis allows us to show how the diffusion of new sexual behaviour starts from "pioneers", before being extended on a community "level" and encountering macrosocietally defined limits, we can detect similar elements in our general population survey.

40The December 1987 survey of the general population of the Paris region recalls some of the observations made two years earlier in the gay population. Although 90% of the respondents had basic knowledge of HIV transmission, beliefs in some transmission possibilities rejected by experts were still widespread. This overestimation of risk transmission in daily life prompted demands for repressive state interventions. Behavioural changes were still a minority phenomenon (Table 7). Multivariate analysis allowed us to distinguish five subpopulations with different social perceptions of AIDS (Figure 3 in Annex II).

41The reactions to the AIDS epidemic in the two extreme and opposed subgroups could be described as "pure" insofar as they reflect a strong internal coherence, not in terms of coincidence with expert knowledge, but in terms of solidified ways of thinking difficult to influence or change. In three other less well-defined subgroups, hesitation prevails on how to react to AIDS.

Table 7: Motivations for condom use

Table 7: Motivations for condom use

* Fear of AIDS and contraception could be cited separately as motivations in that survey.

42The two extremely opposed subgroups with "pure" reactions to AIDS each represents 15% of respondents (Table 8). They can be labelled "coercive" versus "libertarian". The "coercive" group, concentrated in the declining fractions of the middle classes and some parts of the blue collar group, presents the highest risk overestimation in daily situations. They also fear the consequences of AIDS on our social and economic equilibrium. Concentrated in the older age group, they do not feel personally exposed to HIV transmission, have not changed their behaviour, and express clear-cut preferences for coercive measures in the fight against AIDS, including compulsory HIV antibody testing and quarantining HIV carriers and people with AIDS.

43At the opposite end of the spectrum, the "libertarian" group, highly educated single students and professionals in the age group below 35, rejects all state intervention, except information and education activities. Such people often know infected persons personally, and have been tested themselves voluntarily. They form the only group that has already changed its behaviour; in particular, they use condoms if they have multiple partners.

Table 8: Characteristics of "extreme" groups for social perception of AIDS (Paris Region General Population)



Socio-demographic characteristics

Young, highly educated, middle class, single.

Old, low education, declining middle classes and some blue-collar, married.

Risk perception

The ways of contamination are known, the risk can be easily avoided.

Risk exists in everyday situations, risk is not only a health problem but a threat to all social and economic equilibrium.


Feel personally exposed, know HIV carrier. Everybody can get infected, we are all concerned by the risk.

Don't feel personally at risk. We are concerned as a society, as our basic values and future are at stake.

Adaptation to risk

Individual behaviour changes. Mutual support and solidarity.

Call for state intervention. Mandatory testing and quarantine for HIV carriers. Protection of society by exclusion of carriers.

44Three subgroups with mixed reactions to AIDS can be distinguished by the degree of repressive measures they would accept. A first "liberal" subgroup is very closed to the libertarian group, but, living in couple relationships, its members see no reasons for changing their sexual habits. They also have a tendency to accept compulsory testing in a few specific circumstances such as pregnancy, and for groups considered unable to change voluntarily, e. g. prostitutes and IV drug users. The more people have children and feel responsible for them, the more they accept propositions for widespread, general, compulsory testing. The second subgroup with mixed reactions to AIDS consists of people who, without feeling at risk themselves, call for such measures in order to protect their children. To their way of thinking, testing should facilitate a responsible attitude from infected people. The third subgroup, although opposed to quarantine, comes very close to the purely coercive group. In addition to compulsory testing, it accepts - for economic reasons - that social security, insurance companies and employers be informed about the serological status of individuals.

45The analysis of the two extreme groups, the "libertarian" and the "coercive" groups, can improve our understanding of risk adaptation procedures by distinguishing, as we did in the analysis of male homo-and bisexuals, between:

  • individuals' predispositions;

  • factors of the immediate environment (community specific);

  • macrosocietal determinants.

46Personal knowledge of infected people and values of individual freedom (in particular sexual freedom) predispose to individual behaviour changes. These changes are reinforced by a feeling of belonging to the same generation that went through the experience of sexual liberation and by identifying oneself with a single life-style with multiple partner relationships. These individual predispositions and specific immediate environmental factors (generation, lifestyle) are overdetermined by high educational status, youth age and belonging to the intellectual middle classes.

47We found a confirmation of these results when studying the determinants of condom use in heterosexuals with multiple partners.

48In the sexually-active general population of the Paris region, the main predictors of condom use were being single, multiple sexual partners and having been already tested for HIV (Poliak et al. 1989). Table 9 gives the results of a logistic regression model (GLIM procedure) applied to the proportion of condom users in our sample of heterosexuals with multiple partners. Of course, no causal relation must be evoked by these results. But the importance of predictive factors which express individual proximity to the disease (voluntary HIV testing, personal knowledge of a HIV carrier, individual fear of STDs) suggests that it is less risk perception per se than individual implication in social processes dealing with the disease that may influence behaviour. This may also renew the debate on the potential impact of HIV antibody testing on behaviour.

49In contrast, demands for repressive state intervention are favoured by risk overestimation and social distance from infected people. Abstract fears of an epidemic as a threat to our social and economic system seem to confirm deeprooted convictions about the negative social consequences of liberalizing tendencies in our society, even in the absence of the feeling of personal risk exposure. These personal convictions are reinforced, in members of the "coercive" group, by the collective experience of social decline, unemployment and economic fragility in their immediate social environment, experiences that are overdetermined by their low educational status that hinders their ability to adapting to a rapidly changing social environment.

Table 9 - Predictors of condom use in heterosexuals with multiple partners (logistic regression model)

Table 9 - Predictors of condom use in heterosexuals with multiple partners (logistic regression model)

* The confidence interval is not symmetrical because it has been calculated on the log-odds.

50Anticipations about the future of the epidemic also shape these attitudes and reactions. When the "libertarians" advocate individual behaviour changes and mutual support and solidarity whilst refusing administrative measures, they also try to defend their value system, including sexual freedom formed by past experiences, and organized solidarity for their friends and for themselves. The "coercive" attitudes also reflect past experiences of economic fragility and social decline. Unconsciously, AIDS becomes an object of projection for the most varied fears. The repressive state interventions they claim for should protect them not only against AIDS, but against all kinds of social threats.

IV. Conclusions

51Many characteristics of HIV infection create specific difficulties for traditional behavioural models, as follows:

  • the nature and complexity of the threat presented by AIDS is extreme and extends to one's most intimate relationships;

  • extreme uncertainty characterises the entire process from exposure to infection to diagnosis;

  • the social context creates risks of stigmatisation of HIV carriers and AIDS victims.

52It is also obvious that rapid changes in human behaviour are occurring because of the threat of AIDS. Traditional epidemiological risk factors and socio-cultural variables do not seem totally adequate to understand differences of patterns of change in sexual behaviour. We have stressed the necessity of developing analysis at three levels:

  • personal predispositions;

  • immediate environment (community);

  • macrosocietal factors.

53The adaptation of sexual behaviour to the risk of HIV infection is very rapid and unequally distributed according to proximity to the disease, social class, age and education. Therefore, only longitudinal research, starting at an early moment of the process, will allow us to understand the whole cycle and group specific forms and rates of patterns of change. This necessity is emphasised by the evolution of the epidemic itself: whilst the number of sexual partners is the main predictor of infection when incidence is still low, this is not the case when incidence increases, as shown by recent data from studies of American homosexual men (Kingsley et al., 1987).

54Finally, control of HIV transmission requires that changes in behaviour be applied consistently in time and in all types of situations. The factors that contribute to triggering behavioural change are not necessarily the same as those that may favour a lasting change, and instability in risk-reducing behaviour has often been observed.



BANDURA A. (1969), Principles of Behaviour Modification, Holt, Rinehart & Winston, New-York.

BASTIDE S., MOATTI J.P., FAGNANI F„ PAGES J.P. (1989), Risk perception and social acceptability of technologies: the French case, in Risk Analysis, 9: 215-223.

BECKER M.H., JOSEPH J.G. (1988), AIDS and behavioural change to reduce risk: a review, in American Journal of Public Health, 78: 394-410.

DOUGLAS M., WILDAVSKY A. (1982), Risk and Culture: an Essay on the Selection of Technical and Environmental Dangers, University of California Press, Berkeley.

EMMONS C.A., JOSEPH J.G., KESSLER R.C. et al. (1986), Psychological predictors of reported behaviour change in homosexual men at risk for AIDS, in Health Education Quaterly, 13: 331-345.

FISCHHOFF B., LICHTENSTEIN S., SLOVIC P. et al. (1982), Acceptable Risk, Cambridge University Press, New-York.

JANZ N.K., BECKER M.H. (1984), The health belief model: a decade later, in Health Education Quaterly, 11: 1-47.

JETTE A.M., CUMMINGS K.M., BROCK B.M, PHELPS M.C., NAESSENS J. (1981), The structure and reliability of health belief indices, in Health Services Research, 16: 81-98.

JOB R.F.S. (1988), Effective and ineffective use of fear in health promotion campaigns, in American Journal of Public Health, 78: 163-167.

KELLY J.A., ST LAWRENCE J.S. (1988), The AIDS Health Crisis. Psychological and Social Interventions, Plenum Press, New-York and London.

KINGSLEY L., DETELS R., KASLOW R. et al. (1987), Risk factors for seroconversion to human immunodeficiency virus among male homosexuals, in Lancet, 1 (8529): 345-349.

LANCE L. (1975), Human sexuality course socialization: an analysis of changes in sexual attitudes and sexual behaviour, in Journal of Sex Education and Therapy, 2: 8-14.

LEFAURE C., MOATTI J.P. (1983), Les ambiguïtés de l'acceptable: perception des risques et controverses sur les technologies, in Culture Technique, 11: 11-25.

Mc KUSICK L., HORSTMAN W., COATES T.J. (1985), AIDS and sexual behaviour reported by gay men in San Francisco, in American Journal of Public health, 75 (15): 493-496.

MOATTI J.P., MANESSE L., LE GALES C., PAGES J.P., FAGNANI F. (1988), Social perception of AIDS in the general public: a French study, in Health Policy, 9: 1-8.

MOATTI J.P., DAB W„ ABENHAIM L., BASTIDE S. (1989a), Modifications of sexual behaviour related to AIDS: a survey in Paris Region, in Health Policy, 11: 227-231.

MOATTI J.P., LE GALES C„ HENRION H„ PAPIERNIK E. (1989b), Social acceptibility of HIV screening among pregnant women, in Abstract for the Vth International Conference on AIDS, Montreal, June 4-9.

MOATTI J.P., TAVARES J., DURBEC J.P., BAJOS N., MENARD C., SERRAND C. (1989c), Modifications of sexual behaviour due to AIDS in French heterosexual'at risk 'population, in Vth International Conference on AIDS, Montreal, June 4-9.

NELKIN D. (1987), AIDS and the social sciences: a review of useful knowledge and research needs, in Reviews of Infectious Diseases, 9 (5): 980-986.

POLLAK M., SCHILTZ M.A. (1987), Identité sociale et gestion d'un risque de santé, in Actes de la Recherche en Sciences Sociales, 68: 77-101.

POLLAK M. (1988), Les homosexuels et le Sida: sociologie d'une épidémie, A.M. Métaillié, Paris.

POLLAK M., DAB W., MOATTI J.P. (1989), Systèmes de réaction au Sida et action preventive, in Sciences Sociales et Santé, 1:111-135.

PUSKA P., JISSINEN A., TUANILEHTO J. (1985), The community based strategy to prevent coronary heart disease: conclusions from the ten years of the North Karelia project, in Annual Review of Public Health, 6: 147-193.

SHERIF M. (1935), A study of some social factors of perception, in Archives of Psychology, 27, no 187.

SIEGEL K., GRODSKY P.B., HERMAN A. (1986), AIDS risk-reduction guidelines: a review and analysis, in Journal of Community Health, 11 (4): 233-243.

SLOVIC P., FISCHHOFF B„ LICHTENSTEIN S. (1980), Facts and fears: understanding perceived risk, in SCHWEIG C. and ALBERS WA. Eds, Societal Risk Assessment, Plenum Press, New-York.

TURNER M. (1982), Towards a cognitive redefinition of the social group, in TAJFEL H. Ed., Social Identity and Intergroup Relation, Cambridge University Press, Cambridge.

TVERSKY A., KAHNEMAN D. (1974), Judgment under uncertainty: heuristics and biases, in Science, 1985: 1124-1131.


ANNEX I: French surveys on sexual behaviour and HIV risk perception

Data presented in this paper come from a longitudinal study concerning sexual behaviour and its changes in the male homo/bisexual population since 1985 and of detailed surveys conducted in 1987 and 1988, one in the general population and another one including only heterosexual respondents with multiple sexual partners.

1. The methodology of survey in the homosexual community was the following. The difficulties of access to the population, representativity, and reliability have been extensively discussed. Our approach, a yearly survey among the readership of a widely read gay weekly (GAI PIED HEDBO) with a circulation of 20,000 to 30,000 copies, does not guarantee us representativity, but the high number of respondents (1985: 999; 1986: 1200; 1987: 1200; 1988: 1500), the sample stability (variation of less than 2% by age, profession, education, residency) and the size of different subgroups allow satisfactory statistical treatment for trend descriptions and correspondence analysis showing relatively homogenous subpopulations with respect to reactions to AIDS (Poliak & Schiltz, 1987). There is another advantage to this approach: it allows one to shorten the data collection period considerably. Data collection in most research undertaken on homosexuals and using a snowball sampling system takes several months. This is too long for observing changes over time such as the spread of condoms. Our data were collected over a two week period.

Questions included sociodemographic characteristics, sexual behaviour and its changes, acceptance of homosexuality in the family and whole environment, information about AIDS (knowledge about transmission, preferred sources of information, confidence in medical authorities), fear of AIDS and its dimensions, STD precedents, and political measures to combat AIDS. Questions about the test and its results have been added since 1986 (Poliak & Schiltz, 1987; Poliak, 1988).

2. A second study based on personal interviews of a representative sample (n = 900) of the population of 18 and over in the Paris region (quota method for sex, age and occupation with stratification for size of town) was carried out in December 1987. The interview included a description of sociocultural characteristics of respondents and 67 questions on knowledge about the transmission of AIDS, opinions about its prevention, and perceptions of risks for health including risk of HIV infection. The respondents were also asked to answer 12 written questions, in the interviewer's absence, about their sexual behaviour. The response rate to the written questionnaire was high (91%) (Moatti etal., 1988; Poliak et al., 1989; Moatti et al., 1989a). The survey was a cooperative project with the "Observatoire Regional de la Santé" (Regional Centre for Disease Control) of Ile de France (Paris region).

3. A third study was carried out in September 1988 on a national sample (n = 1000) of heterosexuals with multiple partners (individuals reporting more than one sexual partner). The interviewer presented a written card describing various situations of non-monogamous sexual relations in the past six months; people who admitted having been in at least one of these situations were included in the study. Unfortunately, for practical reasons, quotas for sex, age and occupation were defined beforehand. Consequently, this sample cannot be considered representative of the general population of heterosexuals with multiple partners in France. However, it was the first attempt in our country to develop a survey in a large sample of that population (Moatti et al., 1989b). The survey was a cooperative project with the "Comité Français d'Education pour la Santé” (French Committee for Health Education), which was in charge of national public information campaigns on AIDS at the time.

ANNEX II: Figure 3 - Correspondence analysis of answers about AIDS prevention in the general population of the Paris region (December 1987)

Group A


Group B


Group C

liberal without behaviour change

Group D

support mandatory screening

Group E

neo-coercive (without support for quarantine)


1 In the literature on risk perception, such mental processes are often qualified as "biases" to explain the "differences" between "objective" statistical estimation of risk by experts and "subjective" risk, but this approach expresses strong value judgments that we do not share.

2 A fundamental epistemological difference is that expected uility theory, at least in economics, does not pretend to be a model of determinants of real behaviour.

3 CFES-INSERM, N=1000, September 1988.

Table des illustrations

Titre Figure 1: Behavioural changes among French homo/bisexuals
Fichier image/jpeg, 96k
Titre Figure 2: Correspondence analysis of answers to 1985 survey in French homosexual community
Fichier image/jpeg, 168k
Titre Table 7: Motivations for condom use
Légende * Fear of AIDS and contraception could be cited separately as motivations in that survey.
Fichier image/jpeg, 120k
Titre Table 9 - Predictors of condom use in heterosexuals with multiple partners (logistic regression model)
Légende * The confidence interval is not symmetrical because it has been calculated on the log-odds.
Fichier image/jpeg, 128k
Fichier image/jpeg, 84k
Fichier image/jpeg, 92k
Fichier image/jpeg, 100k
Fichier image/jpeg, 72k

© Presses de l’Université Saint-Louis, 1990

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search