American Sexual Behavior: Trends, Socio-Demographic Differences, and Risk Behavior

Tom W. Smith

National Opinion Research Center
University of Chicago

GSS Topical Report No. 25

http://cloud9.norc.uchicago.edu/dlib/t-25.htm

Updated
December, 1998

This research was done for the General Social Survey (GSS) project directed by James A. Davis, Tom W. Smith, and Peter V. Marsden. The GSS is supported by the National Science Foundation, Grant No. SBR-9717727.

Version 3.0

Introduction

Sexual behavior is not only of basic biological importance, but of central social importance. Not only does it perpetuate the human species, but it is the central behavior around which families are formed and defined, a vital aspect of the psychological well-being of individuals, and a component of a variety of social problems. Among current concerns tied in part to sexual behavior are the familial problems of marital harmony and divorce; criminal problems of rape, incest, child molestation, and prostitution; reproductive problems of infertility, sterility, unwanted and mistimed pregnancies, and abortion; and health problems related to sexually transmitted diseases (STDs).

About 17% of adults 18-59 have had an STD and the lifetime infection rate is likely to be over 20% (Laumann, Michael, Gognon, and Stuart, 1994). Moreover, with the advent of AIDS the medical problem of STDs has taken on increasing urgency (Div. of HIV/AIDS Prevention, 1995 and Yankauer, 1994). Deaths from AIDS rose at a rapid pace in the 1980s and early 1990s. By 1992 AIDS had become the number one cause of death among men 25-44. Only recent improvements in medical treatments have curbed the rising levels of HIV to AIDS conversions and lowered the death rate from AIDS (CDC, 1998; "AIDS Falls," 1998; State and Local, 1998). Most HIV infections have resulted from sexual behavior and heterosexual intercourse is increasingly becoming a mode of transmission ("Heterosexuality," 1994; CDC, 1998).

Because of both the importance of sexual behavior in general and the health crisis of AIDS in particular, we need to arm ourselves with a thorough, scientifically reliable understanding of sexual behavior and especially to study high-risk behavior (Hewitt and Beverley, 1996). In this paper we will outline what is currently known about American sexual behavior. Attention will focus on 1) trends and 2) socio-demographic differences within the following areas:

a) Premarital and Adolescent Sexual Activity including Cohabitation and Non-marital Births

b) Adult and General Sexual Behavior including Extra-marital Relations, Gender of Sexual Partners, Frequency of Sexual Intercourse, and Sexual Inactivity

c) The Impact of AIDS on Sexual Behavior including Reported Changes in Sexual Behavior, Number of Sexual Partners, Relationships between Sexual Partners, Prostitution, and the Use of Condoms

Premarital and Adolescent Sexual Activity

Premarital sexual intercourse has become increasingly common over the last century (Table 1A, see also Hopkins, 1998 and Whitbeck, Simons, and Goldberg, 1996). This increase was not merely the result of the so-called sexual revolution of the 1960s. The change was underway for decades prior to the 1960s and has continued since then. Rates among men were moderately high even from the beginning (61% of men born before 1910 report having had sexual intercourse before marriage) and climbed steadily. Women had low rates of premarital intercourse to begin with (only 12% of those born before 1910 had pre-marital sexual intercourse), but their rates grew more rapidly than those of men and the gap between men and women has narrowed over time. By the 1980s (roughly the 1965-1970 birth cohort) women had almost as much sexual experience as men prior to marriage (in 1988 of those 15-19 60% of men and 51.5% of women had engaged in premarital sex). This increase in premarital sexual experience is confirmed by community studies (Wyatt, Peter, and Guthrie, 1988 and Trocki, 1992) and longitudinal panels (Udry, Bauman, and Morris, 1975).

Then in the early 1990s the century-long increase in the level of premarital and adolescent sexual activity reached a peak and then declined for the first time in decades (Table 1A and Bachrach, 1998; Besharov and Gardiner, 1997; Stossel, 1997; and Peipert, et al., 1997). The decrease appears to be greater for males than females, but both genders show a levelling-off and then some reversal.

With the increase in levels of premarital sexual intercourse came a fall in the age of first intercourse (Table 1B). In 1970 5% of women age 15 and 32% age 17 were sexually experienced, by 1988 this had grown to 26% at age 15 and 51% at age 17 (see also Kahn, Kalsbeek, and Hofferth, 1988 and Hofferth, Kahn, and Baldwin, 1987). This trend also may have levelled-off since then, but the evidence is inconclusive (Table 1B and Strunin and Hingson, 1992).

When the increase in levels of premarital sexual intercourse is coupled with the rising age at first marriage, this means that men and women are spending longer and longer periods of their sexual life outside of marriage (Ehrhardt and Wasserheit, 1992; Bachrach and Horn, 1987; Laumann, Gagnon, Michael, and Michaels, 1994). Between 1960 and 1997 the median age at first marriage rose from 22.8 to 26.8 for men and from 20.3 to 25.0 for women. For women the median age of first premarital intercourse in 1960 was about 19.0 (Turner, Miller, and Moses, 1989 and Bachrach and Horn, 1987), meaning on average only a short period of premarital sexual activity. In 1990 the median age at first sex was 16.9 for women (Divs. of Epidemiology and Prevention; Adolescent and School Health; and Reproductive Health, 1992), meaning an average exposure of 8.1 years. For men the period of premarital sexual activity now averages 10.7 years (26.8 - 16.1).

With people spending longer periods engaged in premarital sexual activity the number of lifetime sexual partners has also grown for both men and women (Table 1C). Between the pre-1910 birth cohort and the 1940-49 birth cohort the portion of men with two or more premarital sexual partners rose from 49% to 73%, while for women the gain was from 3% to 26%. This trend continued at least until recent years. For example, among sexually experienced women ages 15-19 living in metropolitan areas 38% had had 2 or more sexual partners in 1971 while by 1988 this had increased to 61%. More recently there is evidence of a reversal of this trend. On the Youth Risk Behavior Surveys (YRBS) the % of male high school students with 4+ sexual partners declined from 1989 to 1997, but the trend among females is unclear.

Cohabitation

The rise in premarital and adolescent sexual activity, coupled with delays in marriage, has led to more people living together. Since 1970 the rate of living together outside of marriage has increased more than 6 fold, from 1.1% to 7.0% of couples (Table 2). Similarly, the proportion of single mothers who were cohabitating grew from 2% in 1970 to 12% in 1995 (London, 1998). While the proportion of couples and adults cohabitating at any one point in time remains small, a large and growing percent live together at some point. Currently over a third of adults in their mid-twenties to mid-thirties cohabited at some point in their lives (Table 3A). Cohabitation after and between marriages is even more common. According to the General Social Survey (GSS) among those 25-44 who are in a second marriage, 61% cohabited with their new spouse before marriage (GSS, 1994).

Cohabitation differs little by gender or race. It is higher among younger adults, the divorced and never married, those in urban areas, and among those who attend church less frequently. Current, but not prior cohabitation with ones spouse, is higher among the less educated and those with lower incomes (Table 3B).

Cohabitation is usually a short-term arrangement, typically resulting in either marriage or a break-up after about a year (median duration of 1.3 years) (Thomson and Colella, 1992; Bumpass and Sweet, 1989; Thornton, 1988).

Cohabitation has often been characterized as a trial marriage and about 40% lead to marriage within two years and about 60% eventually culminate in marriage between the cohabiting partners (Bumpass and Sweet, 1989). However, marriages formed after cohabitation are rated as less stable and result in more divorces than marriages not preceded by living together (Axinn and Thornton, 1992; Brown and Booth, 1996; Clarkberg, Stolzenberg, and Waite, 1995; DeMaris and MacDonald, 1993; DeMaris and Rao, 1992; Lillard, Brien, and Waite, n.d.; Popenoe, 1993; and Thomson and Colella, 1992). Cohabitation thus "does not seem to serve very well the function of a trial marriage... (Popenoe, 1993)."

Those who are cohabiting have fewer sexual partners than those who are unmarried and not cohabitating. However, people who are cohabitating have more sexual partners than married couples (Waite and Joyner, 1996). For example on the 1993-94 GSS the married averaged 0.97 partners last year, the never married who were cohabiting had 1.38 partners, and the non-cohabitating never married had 1.63 partners. That fact coupled with the transitory state of most cohabitations makes living together riskier than marriage when it comes to STDs (Turner, Miller, and Moses, 1989; Kost and Forrest, 1992).

Non-marital Births

With the link between sexual activity and marriage breaking down, the connection between marriage and procreation has also lessened. In the 1960s (and presumably before) when premarital sexual intercourse resulted in conception, the women's pregnancy usually in turn led to a marriage before the child was born (Table 4). Since then, the propensity of unmarried parents to marry before the birth of their child has steadily fallen. By the 1990s less than 25% of women who conceived children before marriage got married before their child's birth.

As a result of the higher level of premarital sexual activity and the decline in marriages after a conception but prior to birth, there has been a large increase in out-of-marriage births (Miller and Heaton, 1991 and Table 5). In 1960 only 5% of all births were to unmarried women. This climbed to 14% by 1975 and 33% by 1994. Then, after over 30 years of increase, the rate leveled-off in 1994-96 at 32-33%.

The trend in the United States has been parallel to changes in culturally similar, advanced industrial nations and not unique to the US. While the percent of births to unmarried mothers climbed from 5% in 1960 to 32% in 1995 in the US, it rose from 5% to 34% in Great Britain, from 4% to 26% in Canada, and from 6% to 37% in France (Statistical Abstract, 1998).

The rate of increase has been much greater for Whites than for Blacks. For Whites the percent of unmarried births has expanded ten-fold from 2.3% of all births in 1960 to 23.7% in 1996, while the Black level grew by just over three-fold from 21.6% in 1960 to 70.4% in 1994. While the Black to White ratio has fallen from a little over 9:1 in 1960 to under 3:1 in the 1990s, the gap between Blacks and Whites has risen from 19 percentage points in 1960 to 44-46 percentage points from 1980 to 1996 (with a peak in 1993). This means that almost a majority of White or Black mothers would have to change their marital status to equal that of the other race. The cumulative difference between Whites and Blacks is further shown by the fact that by ages 30-34 only 23% of never-married, White women have given birth, while 69% of never-married, Black women have had a child (Bachu, 1991 & 1995; Loomis and Landale, 1994).

While both Whites and Blacks have a greater proportion of births occurring outside of marriage, they have achieved the gains through decidedly different paths (Table 5). For Whites the unmarried birth rate (number of births to unmarried women per 1,000 unmarried women ages 15-44) rose throughout the period. It increased more than 4 times from 9 in 1960 to 38 in 1994-95. For Blacks their rate was quite variable over time. It fell from 98 in 1960 to 79 in 1985 before climbing again to 91-93 in 1989-90 - still below their birth rate in the 1960s. In the early-1990s the Black unmarried birth rate then again declined, falling to 76 in 1996.

In addition, there is also a high level of unintended births (Abma et al., 1997 and Williams, 1991). Of women 15-44 in 1995 who have had a child, 28% reported that they had an unintended birth and this reached 36% for women 40-44. Of those with an unintended birth, 80% reported the birth as mistimed and 20% as unwanted.

In brief, over the last century premarital sexual activity has become more widespread, sexual initiation has started at younger ages, the period of premarital sexual activity has lengthened, and the number of premarital sexual partners increased. This expansion in premarital sexual activity in turn led to major increases in cohabitation and unmarried child bearing.

But during the 1990s a small, but historic, reversal of some of these trends occurred. The level of premarital and adolescent sexual activity levelled-off and in some aspects retreated and the proportion of births outside of marriage reached a plateau. These changes are partial rather than across the board (e.g. levels of cohabitation continue to rise) and even those behaviors that have levelled-off or reversed are at near record high rates. But even limited changes to a massive, century-long trend are highly notable and potentially important from a public health perspective.

Adult and General Sexual Behavior

Compared to the amount of information available on premarital and adolescent sexual behavior, until recently there has been little scientifically reliable data on the sexual behavior of adults or of the population in general (Aral, 1994; di Mauro, 1995; and Seidman and Rieder, 1994). Moreover, the dearth of representative and credible studies has created a vacuum that has been filled by unrepresentative and incredible misinformation from popular magazines, sex gurus, and others. In this section we review what is known about extra-marital relations, sexual orientation, the frequency of sexual intercourse, and sexual inactivity.

Extra-marital Relations

There are probably more scientifically worthless "facts" on extra-marital relations than on any other facet of human behavior. Popular magazines (e.g. Redbook, Psychology Today, Cosmopolitan), advice columnists (Dear Abby and Dr. Joyce Brothers), pop- sexologists (e.g. Morton Hunt and Shere Hite) have all conducted or reported on "studies" of extra-marital relations. These studies typically find extremely high level of extra-marital activity (Reinisch, Sanders, Ziemba-Davis, 1988; Smith, 1989; Smith, 1991b; and Gibbs, Hamil, and Magruder-Habib, 1991). Hite for example reported that 70% of women married five or more years "are having sex outside of their marriage (Smith, 1988)." They also often claim that extra-marital relations have become much more common over time. Dr. Brothers (1990), for example, claims that 50% of married women now have sex outside of marriage, double the level of a generation ago.

But representative, scientific surveys (Choi, Catania, and Dolcini, 1994; Forste and Tanfer, 1996; Greeley, 1994; Greeley, Michael, and Smith, 1990; Laumann, Gagnon, Michael, and Michaels, 1994; Leigh, Temple, and Trocki, 1993; Tanfer, 1994; Treas and Giesen, 1996) indicate that extramarital relations are less prevalent than pop and pseudo-scientific accounts contend (Table 6). The best estimates are that about 3-4% of currently married people have a sexual partner besides their spouse in a given year and about 15-17% of ever-married people have had a sexual partner other than their spouse while married (Michael, Laumann, and Gagnon, 1993).

There is little direct and reliable trend information on extra-marital relations before 1988. Since then, levels have not changed. Prior to then there is indirect evidence that extra-marital relations may have increased across recent generations. The figure of ever having extra-marital relations rises from 13% among those 18-29 to 21% among those 40-49 (Table 7). It then falls to 8% among those 70 and older. Since these are lifetime rates, one would normally expect them either to increase across age groups or to increase until a plateau is reached (this would be the case if few first-time, extra-marital relations were started among older adults). The drop among those 50 and older suggests that members of birth cohorts before about 1940 were less likely to engage in extra-marital relations than are spouses from more recent generations (Laumann, Gagnon, Michael, and Michaels, 1994; Greeley, 1994).

In terms of current extra-marital relations Table 7 indicates that they are more common among younger adults. This is largely a function of younger adults having been married a shorter period of time. Some recently married people have difficulty adjusting from a premarital pattern of multiple sexual partners to a monogamous partnership and in general recent marriages are more likely to end in divorce than long-term marriages. The rates of extra-marital relations are about twice as high among husbands as among wives (Table 7). Extra-marital relations are also more common among Blacks, those with lower incomes, those who attend church less frequently, those who have been separated or divorced (including those who have remarried), and those who are unhappy with their marriage. It also may be more frequent among residents of large cities, but the overall relationship with community type is statistically significant only for the last 12 months. Finally, extramarital relations in the last year are more likely to occur among the less educated, but the lifetime pattern with education is mixed and unclear.

Gender of Sexual Partners

Few debates have been so contentious as the controversy over the sexual orientation of Americans (Billy, et al., 1993; Stokes and McKirnan, 1993; Michaels, 1997; and Swann, 1993). The gay and lesbian communities have long adopted 10% as the portion of the population that is homosexual.⁴ However, a series of recent national studies (Table 8A) indicate that only about 2-3% of sexually active men and 1-2% of sexually active women are currently homosexual. These national American estimates are consistent with figures from local communities in the United States (Trocki, 1992; McQuillan, Ezzati-Rice, Siller, Visscher, and Hurley, 1994; Guterbock, 1993; and Rogers and Turner, 1991), indirect measurements (Aguilar and Hardy, 1991), and statistics from Great Britain, France, Norway, and Denmark (AIDS Investigators, 1992; Johnson, Wadsworth, Wellings, Bradshaw, and Field, 1992; Biggar and Melbye, 1992; Melbye and Biggar, 1992; Sundet, et al., 1988; Sandfort, 1998; and Diamond, 1993)(Table 8B).

Rates of same gender contact increase as the reference period is extended. Recent figures (Table 9) indicate that 3.0% of sexually active males have had a male sexual partner in the last 12 months, 3.9% during the last five years, and 5.9% since age 18 (See also Smith, 1991a and Michael, Laumann, and Gagnon, 1993).⁵ As the time frame is lengthened, the % of men with only male partners declines. Over the last 12 months 2.4% are gay and 0.6% are bisexual, over the last five years it is 2.5% gay and 1.4% bisexual, and since age 18 less than 1% are gay and 4%+ bisexual.⁶ Most of those who report both male and female sexual partners since age 18 report only opposite gender partners during the last year (Smith, 1991a). Lesbians follow these same patterns.

There is little reliable evidence on whether sexual orientation has changed before the late 1980s.⁷ In terms of attitudes levels of approval of homosexuality slightly declined from 1973 to 1991, but then rose notably in 1992-98 (Laumann, Gagnon, Michael, and Michaels, 1994; Smith, 1994; and Davis and Smith, 1998). Since then, no change seems to be occurring in the sexual orientation of the public (Table 9).⁸

Studies of male and female homosexuality both in the United States and Europe regularly find a higher proportion of males are gay than the share of females who are lesbian (Tables 8A & 8B and Spira, Bajos, Ducot, 1994; Wells and Sell, 1990; AIDS, 1992; Johnson, Wadsworth, Wellings and Field, 1994; Sandfort, 1998).⁹

Sexual orientation does not very much across socio-demographic groups (Table 9). The one distinctive pattern for both gays and lesbians is that they are less likely to have married. About 60% of those with a same gender partner during the last 12 months have never been married compared to the 16% of female heterosexuals and 21% of male heterosexuals who have never been married.

Second, gays, but not lesbians, are distinctive in congregating in the largest central cities. About 8.5% of men in large central cities have had a same sex partner in the last year as have 9.6% over the last 5 years and 14.7% since age 18. Rates are lowest outside of metropoitan areas.¹⁰ Lesbians, like gays, are underrepresented in non-metropolitan areas.

Third, more gays are found in the lower income categories and among Blacks. Race is unrelated to being lesbian (except weakly for the lifetime figures) and low income is only marginally related to being lesbian. This may partly reflect both homosexual activity in prisons and male, homosexual prostitution. Education does not consistently differentiate among homosexuals.

Fourth, lesbians, but not gays, are more common among younger age groups. This could indicate an increase in homosexual activity among women across cohorts (see also Rogers and Turner, 1991).

Finally, lesbians, but not gays, attend church less than heterosexuals. About 3.4% of women who rarely attend church have had a female sexual partner in the last year compared to only 1.3% of those who attend regularly.

Frequency of Sexual Intercourse

There is some evidence that the frequency of intercourse rose from the 1960s to the 1970s (Trussell and Westoff, 1980) and may have declined in the 1980s. Among teenage males 17-19 living in metropolitan areas the rate fell from 59.8 times per year in 1979 to 39.0 in 1988 (Sonenstein, Pleck, and Ku, 1990, but then among all males ages 17.5-19, it rose from 30 to 49 times per year between 1988 and 1991 (Ku, Sonenstein, and Pleck, 1993). Among unmarried women ages 20-29 the rate showed a more modest decline from 59.8 in 1983 to 56.0 in 1988-93 (Tanfer and Cubbins, 1992 and GSS, 1994). However, no meaningful change has been occurring among all adults since 1988. On average adults engage in sex about 60 times per year, a little over once a week Table 10A).

The overall adult average is relatively uninformative however since the frequency of sexual intercourse varies notably across socio-demographic groups (Table 10B). The factor making the biggest difference is age. Among those 18-29 frequency averages about 84 times per year. This then falls off steadily from 64 times per year for those in their 40s to 9.8 times per year for those 70 and older. Among the married the decline is even more striking, dropping from 112 times per annum for those under 30 to 16 times per annum for those 70 and older. This age related pattern is nearly identical to one shown in the 1988 National Survey of Families and Households (Hughes and Gove, 1992) and is consistent with a large number of other studies (Call, Sprecher, and Schwartz, 1996; Feldman, Goldstein, McKinlay, Hatzichristou, and Krane, 1992; Hawton, Gath, and Day, 1994; Jasso, 1985; Jasso, 1986; Kahn and Udry, 1986; Laumann, Gagnon, Michael, and Michaels, 1994; Leigh, Temple, and Trocki, 1993; Rao and VandenHeuvel, 1995; Tanfer and Cubbins, 1992; Udry, 1980; Udry, Deven, and Coleman, 1982; National Council on the Aging, 1998; and Udry and Morris, 1978).

This decline with age also occurs within marriages. First, the so-called honeymoon effect leads to the highest rates of intercourse among the recently married and those recently married tend to be younger (Greenblat, 1983; James, 1981; James, 1983). Second, biological aging reduces hormonal output and poor health in general and impotency in particular increases with age (Feldman, Goldstein, McKinlay, Hatzichristou, Krane, 1992; Leiblum, 1990; Levy, 1992; McKinlay and Feldman, 1992; Morokoff, 1988; Schiavi, 1990; Schiavi, 1992). As a result, even among couples who rate their marriages as very happy (GSS, 1994) and among those who say they are still "in love" (Greeley, 1991) frequency of intercourse declines with age.

Marital status also influences sexual activity (Table 10B). Frequency is greatest among the currently married (with those remarried slightly exceeding those in their first marriage probably because of the honeymoon effect). The never married and divorced have lower rates, probably because of less continuous and convenient availability of a partner. The widowed have by far the lowest rates, a function of their age as well as their marital status. The edge of the married over the non-married becomes even more apparent when age is taken into consideration. Activity is 25-300% greater among the married compared to the non-married at various ages. Among the married intercourse is more frequent among those who have happier marriages (Smith, 1991a; Waite and Joyner, 1996).

Husbands and wives closely agree on the frequency of intercourse in the aggregate and in most paired comparisons of partners (Bachrach, Evans, Ellison, and Stolley, 1992 and Smith, 1992a and 1992b). However, unmarried men and women differ considerably with men reporting more activity than women do (Bachrach, et al., 1992). This is true even if the surplus of widowed females is accounted for.

A multivariate analysis indicates that more sexual intercourse is separately and independently related to a) being younger, b) having been married less than 3 years, and c) rating ones marriage as happier. It is unrelated to gender with controls for these other factors (GSS, 1996).

There are little differences by race, community type, education, or income and these are mostly related to age and/or marital status.

Sexual Inactivity

Sexual inactivity takes on three distinct forms: 1) the period prior to first sexual intercourse, 2) periods of extended inactivity after first intercourse and prior to last intercourse, and 3) the possible period after last intercourse. The first has been dealt with above in the discussion of premarital sexual relations. The latter two are discussed here. They can not be readily separated from one another with the available data.

Sexual inactivity appears to have modestly declined since the early 1980s (Table 11). Both among women of childbearing age and among all adults the proportion not engaging in sex over extended periods (3-12 months) has decreased in the 1990s.

Among adults there is a u-shaped curve with sexual inactivity most frequent among the youngest and oldest adults. Sexual inactivity among the elderly is fairly common and is a function of general aging, poor health, and unavailability of a partner. As we saw in the section on frequency of sexual intercourse, sexual activity decreases markedly with age even when a partner remains available. This is a function of both a reduction in the rate of sexual intercourse among those remaining sexually active and also an increase in the proportion sexually inactive. Among those over 70 61% are not currently sexually active. Among this age group sexual abstinence reaches 33% among the married and among the non-married it hits 93% (Table 11).

Sexual inactivity is much less common among younger adults. Among the currently married only 1.5-3% of those 18-49 are sexually inactive. Almost all of this group either have poorer than average health and/or rate their marriage as unhappy (Smith, 1992; see also Donnelly, 1993 and Edwards and Booth, 1976). While 7% of married couples of all ages are sexually inactive over the last year (GSS, 1998), fully 16% of married couples have not engaged in sexual intercourse in the last four weeks (Donnelly, 1993, see also Dolcini, et al., 1993). Sexual abstinence is much higher among the non-married ranging between 16% and 27% for those under 50.

Most other socio-demographic differences are small and merely reflect underlying differences in age and/or marital status, but sexual inactivity is lower in households with higher incomes.

While there has been long-term and massive increases in all aspects of premarital and adolescent sexual activity, there is little evidence that similar changes in regards to adult or overall sexual behavior have occurred. Moreover, adult sexual behavior appears to be more restrained and traditional than it has commonly been portrayed.

The Impact of AIDS on Sexual Behavior

AIDS is a deadly and infectious disease that has mainly been transmitted through tainted blood products, sexual intercourse, and the sharing of needles by users of illegal injection drugs. With the safeguarding of the blood supply current transmission occurs largely through sexual intercourse or the sharing of needles with a HIV positive individual. The only means of restricting the spread of the disease is to have people adopt safer sexual and injecting drug use behaviors.

On the one hand, the long latency period of AIDS greatly complicates matters since infected people often are not aware that they are HIV positive and therefore pass the infection on to others. On the other hand, since the mid-1980s over 90% of the public have known that HIV is spread by sexual intercourse and knowledge about AIDS in general has grown over time (Rogers, Singer, and Imperio, 1993 and Singer, Rogers, and Corcoran, 1987).¹¹ Given the existence of widespread, if imperfect, knowledge about the role of sexual intercourse in spreading AIDS, the question arises whether behavior has been modified in light of the known risk.

Reported Changes in Sexual Behavior

A number of studies have asked people whether they have changed their sexual behavior because of AIDS (Table 12) or have taken steps to avoid AIDS (Table 13). Early surveys in 1986-87 showed that only about 7-11% of adults reported any change. At that time these levels were commonly seen as indicating that people were either not informed about the risk of AIDS or were not reacting responsibly to the risk of AIDS. But the recent studies on sexual orientation, extra-marital relations, and sexual abstinence (Tables 6, 9, and 11) indicate that the number of people at risk was smaller than initially feared. And if relatively fewer people were engaged in risky sexual behavior, it would be understandable that few reported altering their behavior. This was directly supported by a 1987 Gallup question in which 68% reported they had not changed their behavior because they were not at risk. Likewise, the low-level of behavioral change among the married (3-12%) compared to the non-married (17.5-51%) reflects the lower level of risky behavior engaged in by married people (Table 12). Similarly, more change is reported by higher risk groups such as younger adults and Blacks.

Of people reporting a change in sexual behavior because of concern about AIDS, about 45-50% mention reducing their number of sexual partners - including having only one partner and getting married, 20-35% cite the use of condoms, 17-30% indicate they have sex less frequently or abstain completely, 10-30% say they are restricting their partners to people they know well, and less than 10% report they have stopped having sex with bisexual men or injection drug users (asked of women only).

Among all adults a number of sexual changes are reported as having been made to avoid AIDS. Monogamy and/or limiting the number of sexual partners is mentioned by about 20%, 10-12% report using condoms, and 5-7% practice abstinence (Table 13).¹²

Reports of behavioral change have risen somewhat over time, apparently indicating that risky sexual behaviors are increasingly being modified (Table 12) and that more people are taking precautions to avoid AIDS (Table 13) (see also Feinleib and Michael, 1998). However, since these questions have not apparently been asked after 1993, it is unknown if this trend continues. Moreover, because of the nature of retrospective questions on behavioral change both the increased trend and the reports themselves are less than ideal.¹³ To reliably track changes due to AIDS, time series monitoring of the relevant risk behaviors are needed. We therefore consider what changes have occurred in sexual behaviors that relate to risk of HIV infection - gay sexual activity, number of partners, familiarity between partners, and condom use.¹⁴

Homosexual Behavior

By the time AIDS was identified, its mode of transmission via sexual intercourse documented, and tests for HIV infection developed, the disease was already widespread among the gay population, especially in San Francisco and New York City. Combined efforts by gay community organizations and public health officials led to the rapid dissemination of knowledge about AIDS and the adoption of safer sex practices by gays. The result was "a dramatic decline in risk practices for HIV transmission...gay men have reduced the number of sex partners, have fewer anonymous sexual encounters, have switched from shorter to longer term relationships, and engaged in less anal intercourse or consistently used condoms (Ehrhardt, Yingling, and Warne, 1991)." Of late however, there has been little further increase in safe sex practices among homosexuals and even some back sliding among some who have tired of the diligence and restrictions required by safer sexual practices, among some minority groups, and among younger gays who did not experience the initial onslaught of the epidemic (Catania, Stone, Binson, and Dolcini, 1995; Ehrhardt, 1992; Ehrhardt, Yingling, and Warne, 1991; Goldbaum, Yu, and Wood, 1996; Kalichman, 1996; Osmond, et al., 1994; Ostrow, Beltran, and Joseph, 1994; Carballo-Dieguez and Dolezal, 1996; and Ostrow, Difranceisco, and Kalichman, n.d.). As a result, same gender sexual intercourse among men remains most frequent mode for the transmission of AIDS ("Update," 1995; Levin, 1995; and State and Local, 1997).¹⁵

Number of Partners

While the overall number of sexual partners among adults has not diminished in recent years (Table 14), some change has been occurring among teenagers and young adults (Table 1C). Among young males the number of partners was probably rising for most of the century until the early 1990s. However, the evidence is somewhat mixed for the 1980s. The mean number of lifetime partners among sexually active males 17-19 in metropolitan areas fell from 7.3 to 6.0 between 1979 and 1988, while among sexually active males ages 17.5-19 the mean number of sexual partners in the last 12 months rose from 2.0 in 1988 to 2.8 in 1991 (Ku, Sonenstein, and Pleck, 1993). During the 1990s there appears to have been a decline in number of partners. The % of male high school students with a lifetime total of 4 or more partners declined from 31% in 1989 to 16% in 1997 (Table 1C and Divs. Adolescent and School Health and Health Interview Statistics, 1994a, 1994b, 1995). For young females there is less clear evidence that the long-term increase in number of partners reversed in 1990s. The YRBS data indicate year-to-year fluctuation rather than any definite trend. However, the GSS shows that from 1988-1990 to 1991-1996 the mean number of sexual partners during the last 12 months for the 18-24 age group significantly decreased from 2.4 to 1.7 (Smith, 1998).

Even if the reductions in number of partners among teenagers is real, many adolescents are still at risk of AIDS and other STDs because of having multiple partners (Anderson and Dahlberg, 1992; Beckman, Harvey, and Tiersky, 1996; Ku, Sonenstein, and Pleck, 1994; Leigh, Temple, and Trocki, 1993; Luster and Small, 1994; Smith, 1991; Trocki, 1992; and Tubman, Windle, and Windle, 1996).

Whether the possible decline in number of partners accumulated by teenagers and young adults will translate into a lower number of lifetime sexual partners is unknown. If it does, it will reverse an expansion that began several generations ago. We can see evidence of that rise in the figures on number of sexual partners since age 18 (Table 14). The increase in the number of sexual partners from ages 18-29 to 40-59 mostly represents the accumulation of partners over ones lifetime. The sharp drop in cumulative partners for those 60 and older occurs because this age group represents a generation that came to age before the peak in premarital sexual activity described above. That is, this generation had fewer premarital partners, married relatively early, and, as a result, has accumulated a lower number of sexual partners than subsequent generations.

Among adults, having multiple sexual partners during the last year and during the last five years is most strongly associated with being young, unmarried, and male.¹⁶ It is also higher among Blacks, residents of large central cities, those with low incomes and less education, and infrequent church attenders. The adult lifetime figures show a similar pattern except that there is no relationship between income or race and number of sexual partners and the less educated have fewer partners than the better educated. The reversal of the education relations results from less educated, earlier cohorts having fewer partners than more recent and better educated cohorts have had.

Multiple partners are thus found in two main social niches, among young, unmarried adults and adolescents who have not yet "settled down" and among disadvantaged segments of society in general and among inner-city minorities in particular who also tend to lead less stable and conventionally-ordered lives (Ford and Norris, 1995 and Wagstaff, et al., 1995).

Relationship to Sexual Partners

Risk increases not only with one's number of sexual partners, but also with the casualness and transitoriness of relationships. When it comes to STDs one "sleeps not only with a partner, but with all of that partner's partners." Closer relationships are associated with (but do not guarantee) mutual monogamy, while casual relationships come without any likelihood of exclusivity.

The trends in relationships are mixed and depend on the measure and data set being examined. First, since 1988, the GSS item on relationships to sex partners during the last year shows little change in relationships between sexual partners either among all adults or among unmarried people under 40 (Table 15A). Most people are engaged in close and presumably mutually monogamous relationships as spouses or cohabiting partners, but each year 3-4% of sexual partners are pick-ups, one-night stands, prostitutes (see below), or other casual couplings. In addition, another 4-5% of partners are better known (neighbors, co-workers, long-term acquaintances), but are not considered close friends or regular partners. Second, there was a small, but statistically significant, drop in whether one was in a continuing relationship with ones most recent sexual partner from 92% in 1996 to 90% in 1998 (Table 15A). Finally, across birth cohorts or women relationship with their first sexual partner have become more casual over time (Table 15A). Of those born in 1951-55, 32% were engaged or married to their first sexual partner, 51% were going steady, 16% were less closely involved, and 1% were in other relationship. For those born in 1976-1980, 4% were engaged or married, 73% were going steady, and 23% were less connected.

More casual relationships (pick-ups, prostitutes, and acquaintances) are most prevalent among the young, unmarried, and males. They are also more common among Blacks, residents of large central cities, and those with lower incomes. Similarly, having ones last sexual encounter with someone that one did not have an "on-going relationship" with is more common among men, Blacks, the young, never married, city residents, the less educated, and infrequent church attenders (Table 15). Household income is not related to non-relational sexual encounters. "One-night stands" are equally common for Black and White males, but less frequent for Black females than for White females (Tanfer, 1994). In general, we see that those socio-demographic groups with a high number of partners also tend to have less familiar partners.

Prostitution

At a time when prostitution could be a major avenue for the spread of AIDS into uninfected areas and groups, we know little about its magnitude or how the situation has been changing. As the National Academy of Sciences' Committee on AIDS Research observed, "(I)nformation about women who work as prostitutes is scant, and knowledge of their clients is sketchier still (Miller, Turner, and Moses, 1990; see also Turner, Miller, and Moses, 1989)."

Because prostitution is an illegal (except in rural Nevada) and socially stigmatized occupation, the amount of reliable information on it is limited. The only time series data come from the arrest records compiled by the FBI (Table 16). They show that the arrest rate climbed from around 30 per 100,000 total population in the early 1970s to a high of 59 in 1982 and 1983. The rate then fell back to the lower 40s by 1988 and has remained at that level since then. Whether this represents shifts in the prevalence of prostitution or variations in law enforcement efforts is unknown.

While the illegality of prostitution is probably the main barrier to accurate counts, estimates are also complicated by the prevalence of many part timers, the continual occupational turnover, and the apparent geographic mobility of prostitutes. There are some more recent and limited studies that have tried to overcome these serious problems and either to estimate the number of prostitutes or of certain types of prostitutes in given localities (Potterat, Woodhouse, Muth, and Muth, 1990; Kanouse, Berry, Duan, Lever, and Richards, 1991; and Leyland, Bernard, McKeganey, 1992) or to measure the proportion of women who have engaged in sex for pay (McQuillan and Ezzati-Rice, Siller, Visscher, and Hurley, 1994, Wyatt, Peters, and Guthrie, 1988; and Brunswick, et al., 1993). In addition, there are recent estimates of what proportion of men have engaged in paid sex.

Although all three of the community, aggregate-estimate studies were carefully done and show a high degree of consistency, all estimates of the number of prostitutes are fraught with uncertainties since they deal with what one study aptly calls a "covert" population. In addition, the Los Angeles and Glasgow estimates are by definition incomplete because they cover only street prostitutes.

Surveys of general populations of women are equally limited. The studies cover sub-groups in local areas, have very small to medium sample sizes (LA=120, Harlem Panel=187, Dallas=745), use different measures, and show different levels of involvement in paid sex.

In addition, two recent samples provide the first national estimates of the proportion of women involved in paid sex. A 1991 survey of 1,669 women ages 20-37 found that 2.0.% had ever had "oral, anal, or vaginal sex in exchange for money or drugs" (Tanfer, 1994) and the 1991-98 GSS of 5,700 women 18+ indicated that 1.6% of women had "had sex with a person you paid or who paid you for sex" since age 18 (GSS, 1998).

Given the differences in ages and measurements, the Dallas survey and the two national samples are in close agreement. They suggest a much higher rate of female participation than the aggregate counts (on the order of 15-20 times higher), but the two sets of estimates are not directly comparable. (Without information on duration, level of involvement, and related factors the survey estimates can not be converted into point estimates of women engaged in prostitution nor can they separate out occasional participants from full-time professionals.)

In brief, the available studies are extremely limited in number and most are unrepresentative of the United States as a whole (one study of course is not even from the United States). In particular, extrapolations from these few local studies to national estimates could well be wrong, especially if prostitution is heavily concentrated in urban centers. This possibility is supported by the fact that on the 1988-1998 GSSs 0.3% of men living in rural areas reported having sex with a prostitute during the last year, while 2.0% of those living in the 12 largest central cities reported having sex with a prostitute during the last 12 months (Table 17). In addition, the lifetime figures show a similar pattern.

Reports by men on paying for sex indicate that 0.6% of men had a prostitute for a sex partner during the last year (GSS, 1998), 5.9% within the last five years (Wells and Sell, 1990), and 16.3% at some point during the past (GSS, 1998; see also Rubin, 1990; McQuillan, Ezzati-Rice, Siller, Visscher, and Hurley, 1984).¹⁷ Unfortunately, these figures are not consistent. The five-year figures are more than twice what would be expected based on the annual figures.

In addition, comparing the annual rates to estimates of FTE prostitutes (assuming that the above urban rates can be applied nationally) comes to only 9.9 clients per prostitute.¹⁸ Thus, if the estimates of number of prostitutes are correct, this would suggest that men are underreporting their number of paid sex partners (either by not reporting partners who were prostitutes or reporting them as falling in another category such as casual dates or acquaintances). Alternatively, the number of FTE prostitutes may be overestimated.

Based on the analysis of reported contact with prostitutes during the last year and during one's lifetime (Table 17), sexual activity with prostitutes does not consistently vary by education or age. As one would expect, lifetime contact generally increases with age, but current use is unrelated to age. Lifetime contact is unrelated to education and current use has an irregular relationship. Contact is higher among those living in metropolitan areas, Blacks, those with lower incomes, veterans (probably when in military service), those who attend church less frequently, and those having gone through a divorce (it appears to be especially high among separated men). Among married men paying for sex during the last 12 months is strongly related to low marital happiness.

Use of Condoms

Undoubtedly because of the advent of AIDS and the dissemination of safer sex messages, condom use doubled from the late 1970s/early 1980s to the late 1980s (Table 18) (see also Moran et al., 1990 and Douglas, et al., 1997). From the mid-1960s to the mid-1970s just over a fifth of women used a condom at the time of their first intercourse. By the mid-1980s this had almost doubled to 42%. Similarly, in 1979 21% of teenage males reported using a condom at the time of their most recent intercourse and in 1988 the level increased to 57.5%.¹⁹ Condom use has continued to increase since then (Beckman, Harvey, and Tiersky, 1996; Catania, Binson, Dolcini, Stall, Choi, Pollack, Hudes, Canchola, Phillips, Moskowitz, and Coates, 1995; Catania, Coates, Peterson, Dolcini, Kegles, Siegel, Golden, and Fullilove, 1993; Catania, Stone, Binson, and Dolcini, 1995; Ford and Norris, 1995; Ku, Sonenstein, and Pleck, 1994; Moore, et al., 1992; Division of Epidemiology and Prevention Research, et al., 1992; Pleck, Sonenstein, and Ku, 1993; Ku, Sonenstein, and Pleck, 1993; Strunin and Hingson, 1992; Piccinino and Mosher, 1998; and Peipert, Domagalski, Boardman, Daamen, McCormack, and Zinner, 1997). For example, the YRBS shows that condom use steadily increased for both males and females in the 1990s and condom use at last intercourse among never married males 15-19 grew by 10 percentage points from 1988 to 1995 (Table 18).

However, while condom use has grown appreciably, it is still far below the general and consistent use called for by safer sex practices (Kost and Forrest, 1992; Pleck, Sonenstein, and Ku, 1991; Potter and Anderson, 1993; Leigh, Morrison, Trocki, and Temple, 1994; Peterson, Catania, Dolcini, and Faigeles, 1993; Sabogal, Faigeles, and Catania, 1993; Grinstead, Faigeles, Binson, and Eversley, 1993; Catania, Coates, Golden, Dolcini, Peterson, Kegeles, Siegel, and Fullilove, 1994; Nguyet, Maheux, Beland, and Pica, 1994; Binson, Dolcini, Pollack, and Catania, 1993 and Douglas, et al., 1997). Among sexually experienced college students in 1995 only 38% reported always using a condom (Douglas, Collins, et al., 1997). Likewise, a 1991 national survey of men 20-39 found that only 26.5% of sexually active men had used a condom during the last four weeks and even among unmarried men with no regular sexual partner only 46% had used a condom during the prior month (Tanfer, Grady, Klepinger, and Billy, 1993, see also, Grady, Klepinger, Billy, and Tanfer, 1993 and Catania, et al., 1992). Similarly, among unmarried women 15-44 in 1990 with 2+ partners in the last 3 months only 16% always used condoms and 39% never did (Mosher and Pratt, 1993). Also, among both men and women 18-24 in 1996 whose most recent sexual partner was not someone they were in an ongoing relationship with only 56% had used a condom (Smith, 1998).

Condom use is higher among socio-demographic groups that have multiple, sexual partners and less committed and on-going relationships with sexual partners. Condoms are used more frequently by Blacks, the young and never married, residents of large cities, those with lower incomes, and those who attend church less regularly (Table 19). Use generally increases with education, but drops off among those with graduate degrees.

Men tend to report greater condom use than do women (especially among teens), but both men and women agree on the trends and general patterns reported here (Divs. of Reproductive Health and Adolescent and School Health, 1992; Leigh, Temple, and Trocki, 1993; Marin, Gomez, and Hearst, 1993; Laumann, Gagnon, Michael, and Michaels, 1994; Leigh, Morrison, Trocki, and Temple, 1994; Douglas, Collins, et al., 1997; and Santelli, et al., 1997). Among adults 21% of men and 18% of women reported using a condom the most recent time they had sex.

Condom use declines among youths and young adults as they age (Table 19; Sonenstein and Pleck, 1997 and Reitman, et al., 1996). This is believed to be largely because sexual relationships become more established and enduring. However, this connection has not been well-established in the studies to date.

Condom use also varies by status of sexual partner. A 1991 study of Hispanic adults and of non-Hispanic White adults living near Hispanics in the Northeast and Southwest indicated that even among those with two or more sexual partners only about 50% always used a condom with their secondary sexual partner(s) (Marin, Gomez, and Hearst, 1993). In addition, condom use is often notably lower among primary partners (i.e. spouse, cohabiting partner, or regular sexual partner) than among secondary partners (Leigh, Temple, and Trocki, 1993; Marin, Gomez, and Hearst, 1993; Ehrhardt, Yingling, and Warne, 1991; Dolcini, et al., 1993; Lansky, Thomas, and Earp, 1998; Albert, Warner, and Hatcher, 1998; Rietmeijer, et al., 1998; and Miller, Turner, and Moses, 1990. But in contrast see Soskolne, Aral, Magder, Reed, and Bowen, 1987). This increases the chance of spreading AIDS and STDs to one's primary sexual partner.

Summary

Since early in this century the bonds between marriage and sexual activity have been unravelling. More men and women have engaged in premarital sexual intercourse, they have become sexually active at earlier ages, and they have accumulated more sexual partners. While premarital and adolescent sexual activity has grown for both men and women, the largest change has been in the sexual behavior of women. The expansion of sexual behavior has in turn led to a rise in cohabitation and a surge in non-married births, and contributed to the growth of various public health and social welfare problems (Besharov and Gardiner, 1993).

Rather than being an isolated phenomenon these changes in sexual behavior, living together, and child bearing have been part of broader social changes towards an individualistic rather than a family-center society (Glenn, 1987; Popenoe, 1993; and Smith, 1997) and towards modern rather than traditional roles for women (Firebaugh, 1990 and Simon and Landis, 1989). Moreover, there are suggestive signs that parallel shifts have occurred in other post-industrial societies. As such, the changes in American premarital and adolescent sexual behavior may result from the development of advanced economies, welfare states, and liberal governments in general rather than from any special situation peculiar to America.²⁰

Of late however this long-term trend has moderated and in a few limited, but key, aspects reversed. First, the increase in premarital and adolescent sexual activity has ended and to some degree has waned. So far this development does not seem to have received much recognition in the mass media or among the public (Stodghill, 1998; Smith, 1998). Second, the portion of nonmarital births has levelled-off (albeit at near record levels). Third, condom use more than doubled during the last 20 years and apparently continues to grow. While there have not been decreases in all forms of risky sexual behavior in all segments of the population, these departures from the long-term trend are notable and may reflect an underlying, nascent shift in social values.

While marriage is no longer the portal into sexual activity for many Americans, it remains an important regulator of sexual behavior and thus a barrier to AIDS and other STDs. Since most married people most of the time engage in sex only with their marriage partner, marriage limits one's total number of sexual partners and reduces the spread of HIV. However, marriage may be less of a barrier than it used to be. The decline in reported rates of ever having had extra-marital relations among those 50 and over does suggest that monogamy may have declined across recent generations. But, on the other hand, there has been no decrease in disapproval of extra-marital relations (Smith, 1990; 1994 and Davis and Smith, 1998), extra-marital relations have not increased since 1988, and "affairs" are much less common than presentations in either pop and pseudo-scientific studies or the entertainment media suggest.

Of course, marriages themselves are also not as enduring as they used to be. The two-and-a-half fold growth of the divorce rate from the 1960s to the early 1980s and its continuation at near historically high levels to the present means that over half of all recent marriages will end in divorce. For most divorced people this means accumulating new sexual partners and especially for those under 50 this often means having multiple sexual partners (Stack, 1992).

Besides marital status sexual behavior is strongly influenced by age. In general, sexual activity diminishes with age with fewer people having multiple partners, less extra-marital sex, frequency of intercourse declining, and sexual abstinence increasing. Cohabitation rates also fall and non-marital births decline with age (ceasing of course for women after menopause).

There are also large differences between Whites and Blacks in their sexual behaviors (Bowser, 1992; Brewster, 1994; Sterk-Elifson, 1992; Kilmarx, et al., 1997; Peterson, Catania, Dolcini, and Faigeles, 1993; Brunswick, et al., 1993; Reitman, et al., 1996; and Quadagno, et al., 1998; but see Wyatt, 1989). Blacks become sexually active at an earlier age, accumulate more sexual partners over their lifetime, have more casual partners, are less likely to marry, have less stable and shorter-term marriages, and have many more children born outside of marriage. Black sexual and child-bearing behavior puts African-Americans at greater risk of contracting AIDS and other STDs (and Blacks do have higher HIV and STD infection rates) and contributes to such problems as single-parent families and childhood poverty.

Sexual behavior also varies by community type. Residents of large central cities have more sexual partners, more casual partners (including prostitutes), and more extra-marital relations than those living in rural areas. In addition, probably due to selective migration, gays concentrate in large cities. Overall since risk behaviors (both sexual and injection drug use) are more common in large cities and the HIV virus is more prevalent in these localities, the chances of becoming infected is especially high in large metropolitan areas (Catania, et al., 1992).

Finally, religion exercises a traditional restraint on sexual behavior (Thornton and Camburn, 1989; Seidman, Mosher, and Aral, 1992; Stack and Gundlach, 1992; Tanfer and Schoorl, 1992; and Goldscheider and Mosher, 1991; Hogan, Sun, and Cornwell, 1998; and Brewster, Cooksey, Guilkey, and Rindfuss, 1998). Those who attend church regularly are less likely to a) become sexually active, b) have multiple and casual partners, and c) among the married, have sexual partners other than their spouse. Church attendance, like rural residence, imposes traditional restraint on sexual behavior.

Given the deadly nature of AIDS, the near universal knowledge of the disease, and the widespread understanding that it is transmitted through sexual intercourse, its impact on sexual behavior has been limited. The largest changes occured among gays in large metropolitan centers who adopted considerably safer sexual practices. But the on-going spread of AIDS from male-with-male sexual contact indicates the continuing shortcomings in safer sex practices among gays.

Among the heterosexual population the largest change has been the increased use of condoms. However, condom use is incomplete and haphazard with condoms being used much less consistently than called for by safer sex standards. In addition, the small decreases in the number of partners among adolescents and youths may also result from the AIDS epidemic. But most people still have numerous premarital sexual partners and many sexual partners represent casual and short-term relationships. Moreover, it is unclear whether the somewhat moderated number of teens and young adults involved with multiple partners will lead to a reduction in the lifetime number of partners. The continuingly high level with multiple partners and the sporadic, if improved, use of condoms means that millions continue to expose themselves each year to the risk of AIDS and other STDs (Smith, 1991b; Anderson and Dahlberg, 1992; and Dolcini et al., 1993). In addition, the level of non-married births remains at near-record levels and the % of all births that are unplanned, also remains high.

In sum, contemporary patterns of sexual behavior are a source of considerable public policy concern relating to AIDS and STDs, child-bearing and child- raising, and many other social problems.

End of Part 1

Part 2 begins here

American Sexual Behavior: Trends, Socio-Demographic Differences, and Risk Behavior-part 2