French Prime Minister Edith Cresson was quoted in Newsweek (Marshall, 1991) regarding her perception of a lack of male interest in women among non-Frenchmen: “The majority of those men are homosexual — perhaps not the majority—but in the U.S.A. there are already 25 percent of them and in England and Germany it is much the same. You cannot imagine it in the history of France.” She did concede that “it’s difficult to produce a statistic” on comparative homosexuality. To this the Newsweek writer editorially contributed: “Evidence suggests that about 10 percent of any country's men are homosexual” (Marshall, 1991). Reference to such figures are common, frequent, and worldwide (e.g., Yamaguchi, 1991; Gargon, 1991).

Writers in scientific literature, unfortunately, are not always better than politicians or journalists. Deisher (1989) in the preface to Gay and Lesbian Youth purports, without reference: “in the United States alone there are 30 million young people between the ages of 10 and 20, 10% of whom are felt to be predominantly or exclusively homosexual.”

A best guess for the origin of these numbers is that they are an extrapolation from the original work of Kinsey et al. (1948, 1953). These works are deservedly considered monumental in their scope and importance. Nevertheless, these studies were not based on a random sample so extrapolations from the data must be made with caution. These numbers are of more than academic or social interest. They are a focus of political debate where the gay population would have them larger and homophobic forces would have them smaller (Buchanan, 1990). And most crucially, they are of scientific importance in model construction and epidemiological forecasting when AIDS transmission is considered a function of different types of homosexual, bisexual and heterosexual practices and the estimated size of these groups in the population (Coolfont Report, 1986; Dietz, 1988; Schwager et al., 1989; Catania et al., 1990).

Despite the AIDS epidemic as a prod to such research, attempts at a national United States study to obtain accurate data have been stymied by political and sexophobic considerations (Booth, 1989). Attempts to similarly frustrate such research in Great Britain were successful for a period but overcome (Wellings et al., 1990). Without specific and new data, in the U.S., attempts to obtain the best estimates for such figures have been made. Fay et al. (1989) reassessed the original data of Kinsey et al., now more than 40 years old, in light of new sophisticated analysis techniques. They also utilized data from a 1970 Kinsey Institute-sponsored study conducted by the National Opinion Research Center (NORC), and a 1988 General Social Survey of leisure time activities reported by Michael et al. (1988). Rogers and Turner (1991) have also used these data and some others to estimate male-male contacts. The findings from both studies as well as others are discussed below. Many of the studies reviewed are recent, have not yet been reported, and are of males and females, and include populations in Europe and Asia not widely known in the U.S. A call for just such data has been issued (Aldhous, 1992).


Before presenting the data themselves it is prudent to first comment on some methodology differences among the studies. All were not measuring the same thing. Each has its own nuances in approach and interest which affects the nature of the data collected and any subsequent analysis. The Kinsey studies are used as a basis for comparison. Particulars of the methods used by the different studies are indicated in Table 1.

Table I. Homosexual and Bisexual Activity Among Different Populations

Reference and questiona Methodb Age
Pop. size Homosexualc
Male %
Female %
Davis and Smith, 1988 (U.S.)
Q. Since your 18th birthday … how many male partners have you ever had sex with?
inter w/ quest
>18   2.4 (K 1-6)  
Diamond et al. [US]
Q. What is your sexual orientation? Do you usually consider yourself:
het/usually het/bi/usually hm/hm
Statewide; residential;
used 5-position Kinsey scale
>18 1024 male
990 female
1.7 (K = 6)
1.9 (K = 5-6)
1.1 (K = 1-4)
0.2 (K = 6)
0.3 (K = 5-6)
0.9 (K = 1-4)
Dixon et al., 1991 [U.S.]
Q. Any sex with another male 1978-1989?
inter w/ quest
18-54 701 male
745 female
73 (K = 1-6)
Fay et al., 1989 [U.S.]
Q. Age at last same-sex experience?
(analyzed for all those after 21)
197o NORC.
secondary analysis of previous surveys
>21 1450 male 6.7 (K = 1-6) N.G.
Harris Poll, 1988 [U.S.]
Q. Any same-sex partner
  previous five years?
  last year?
  last month?
H.S. inter w/quest
same-sex interviewer
16-50 739 male
409 female

4.4 (K = 1-6)
3.8 (K = 1-6)
1.8 (K = 1-6)

3.6 (K = 1-6)
2.8 (K = 1-6)
2.1 (K = 1-6)
Hart, 1968 [Philippines] Anthropological field work Postpubertal 729 persons
(total population)
1.6 (K = 1-6) 0
Asayama. 1976 [Japan]
Q. Any homosexual contact such as kissing, petting or mutual masturbation?
Students (H.S. & Univ.); questionnaire
16-21 2574 male
2101 female
7.1 (K = 1-6) 4.0
JASE [Japan]
Q. Ever any homosexual body contact?
1975 10-22 2764 male
2236 female
6.7 (K = 1.-6) 3.8 (K = 1.6)
1981 10-22 2505 male
2485 female
5.8 (K = 1.-6) 5.0 (K = 1.-6)
Randomized students (HS. & Univ.)
10-22 4317 male 3.7 (K = 1.-6) 3.1 (K = 1.-6)
Michael et al.., 1988 [U.S.]
Q. Any same-sex experience last 12 mos?
Analysis of
NORC 1988 survey
>18 504 male
567 female
2.8 (K = 6)
0.4 (K = 1-5)
0.2 (K = 6)
0.0 (K = 1-5)
Morens and Pollot, in prep. [Palau]
Q. What types sex experience last 12 mos? Hetero/homo/bi
random 10% of population
20-90+ 424 male
422 female
1.9 (hm)
2.8 (bi)
2.8 (hm)
0.7 (bi)
Rogers and Turner, 1991 [U.S.] Analysis of previous NORC surveys and Dixon et al. survey; inter w/ quest     K = 1-6 N.G.
1970 Kinsey >21 3018 male 6.7  
1988 NORC >18 1481 male N.G.  
1989 NORC >18 1537 male 4.9  
1990 NORC >18 1372 male 4.8  
1989 Dixon et al. 18-54 1449 male 8.1  
Schover and Jensen, 1988 [Denmark]
Q. Any same-sex experience during adulthood?
Interview of women by female physician during health prevention study 22-70 625 female   K = 1-6
< 0.2
Sittitrai et al.., 1992 [Thai]
Q. Any same-sex experience during adulthood?
het/usually het/bi/usually hm/hm
used 5 position Kinsey scale
Postpuberty 983 male
1285 female
0.3 (K = 6)
3.4 (K = 1-6)
0.9 (K = 6)
1.2 (K = 1-6)
Smith, 1991 [U.S.]
Q. Since your 18th birthday … how many male partners have you ever had sex with?
inter w/ quest
>18 1401 0.7 (K = 6)
5.6 (K = 1-6)
Wellings et al., 1990 [U.K.]
Q. Any same-sex experience?
national random survey
inter w/ quest
16-59 1000 male
and female
9.0 (K = 1-6) 4.0 (K = 1-6)
Zessen and Sandfort, 1991 [Dutch]
Q. Sexual behavior past 12 months?
used 7-position Kinsey scale
18-50 421 male
580 female
3.3 (K = 5-6)d
8.3 (K = 1-5)d
0.4 (K = 5-6)d
6.9 (K = 1-5)d
Q. Ever in life had same-sex sex?   12.0 (K = 1-6) 4.0 (K = 1-6)
a Q. This is the essence of the question wording.
b H.S. = Random/Probability Household Survey; inter w/ quest = personal interview with sensitive questions responded to by self-administered questionnaire; NORC = National Opinion Research Center Study; RDD = Random digit dialing.
c K*= my interpretation of the authors description or the authors designation; N.G. = not given/not asked/not reported; hm = homosexual; bi = bisexual.
d “Sexual preference” regardless of behavior during preceding 12 months.

Kinsey et al. (1948, 1953) developed a 7-point scale (0-6) with which to characterize their respondent’s behavior. This essentially offers a rough comparison of same-sex to opposite-sex encounters. To further qualify the activity, Kinsey et al. also asked about fantasy behavior. Each individual was thus identified with separate ratings for behavior and fantasy. Typically then, these two ratings for the individual were combined for a composite classification of both overt behavior and psychological response. Few of the post-Kinsey general population surveys obtained a Kinsey rating for their respondents or asked about fantasies (Table I). Studies of specific gay populations, however, generally did use both scales (Table II).

Table II. Kinsey Ratings of Behavior and Fantasy Among Males Self-Identifying as Gay (Percentages)

Kinsey scale Weinberg and Williams, 1974a
(N = 239)
Bell and Weinberg, 1978b
(N = 575)
McWhirter and Mattison, 1984c
(N = 312)
Higa, 1988d
(N = 276)
6 51 74 82 77
5 30 18 11 19
4 13 3 7 2
3 4 5 0 2
2 2 0 0 0
1 0 0 0 0
0 0 0 0 0
6 NR 58 76 64
5 NR 28 16 25
4 NR 8 9 8
3 NR 6 0 3
2 NR 0 0 0
1 NR 0 0 0
0 NR 0 0 0
a Gay white U.S. males. NR = not recorded
b Gay white U.S. males.
c Committed gay male U.S. couples.
d Gay multiracial males in Hawai‘i.

The Kinsey scale had obvious advantages. First, it offered a quantifiable description of a group of acts and thoughts. And since behaviors for many individuals shift along a life course, with further refinement Kinsey et al. were also able to describe such changes.  They recorded prepubertal and postpubertal behaviors that were premarital, marital, and postmarital in 3-year groupings. Second, the scale allowed investigators to describe the respondent's activities regarding the sex of his or her partners.  The respondents did not have to identify “homosexual” or “bisexual” if they considered these labels stigmatizing. Indeed, compared with research evaluations by their predecessors, use of this scale offered a marked leap forward.

But, in the ensuing years, thinking on this issue as became more sophisticated. Would individuals who had 1 or 2 homosexual experiences and 100 heterosexual ones be rated the same as someone who has had 1 and 10? Would they both be a K = 1 or both be considered bisexual? Do we include or exclude adolescent experimentation or jail or prison experiences that are never repeated? How should one distinguish between self-identification and labeling by professionals? Should we reserve the terms “homosexual, ” “bisexual, ” or “heterosexual” for an individual’s activities or fantasies or both and should we only consider acts that include orgasm? The surveys reviewed were not uniform in this regard. Even the basic question on same-sex behavior varied among studies. A general question might ask something like “Have you ever participated in same-sex activity.” Some surveys from Japan (Japanese Association for Sex Education: 1975, 1981, 1987) ask of “any homosexual body contact.” The implications of all of these is most probably similar but the clarity of the question sometimes leaves room for interpretation. Most, however, would probably be casting a wide net. Only two (Michael et al., 1988; Fay et al., 1989) followed the Kinsey model of asking of same-sex contact when either participant came to climax.

The studies reported here differ greatly in the periods they consider of interest. Some ask of behavior since puberty, others since 10 or 16 or 18 or 21 years of age, others “ever in your life, ” some ask of the last 12 months, and one (Dixon et al., 1991) concentrates on the HIV crucial period 1978-1989 (see Table I). Some few have upper cutoff ages, e.g., 54 by Dixon et al. (1991) and 70 by Schover and Jensen (1988), while a couple of studies report exclusively on college students.

Some researchers just ascertained if there were both same-sex and opposite-sex encounters in the respondent’s history (e.g., Rogers and Turner, 1991). Others ask the respondents to self-identify using verbal expressions of the 7-point Kinsey scale, a modified 5-point scale, or ask if the individual sees himself/herself as heterosexual, homosexual, or bisexual or had demonstrated such behavior (see Table I). But these terms themselves are not always defined clearly. Kinsey et al. (1948, 1953) considered ambisexuals2 as those demonstrating K = 1 through K = 5, Weinberg and Williams (1974) and Bell and colleagues (Bell and Weinberg, 1978; Bell et al., 1981),  as did Green (1987), considered bisexuals as those whose behaviors and fantasies averaged K = 2 through K = 4. Haeberle (1978), while recognizing the overlap, categorized heterosexuals as K = 0-2, homosexuals as K =4-6, and ambisexuals as K =1-5. All these investigators were coping with the reality that the terms often obscure as much if not more than they reveal.

It is rare that an individual would both live and fantasize an erotic life with someone who was not sexually arousing, but an occasional or even prolonged “fling” might occur. Indeed the work of many document this (e.g., Kinsey et al., 1948, 1953; Rogers and Turner, 1991). Necessity might also force such behavior. In addition, any individual may exhibit heterosexual, homosexual, or bisexual behaviors for many reasons that have nothing to do with sexual arousal or interest. An individual, for example, may engage in homosexual or heterosexual activities for friendship, due to coercion or curiosity, or a host of other reasons (Diamond, 1992). While Kinsey et al. tried to probe motivation, none of the studies reviewed here did.

Motivation seems particularly at issue in considering bisexual activity. What percentage of the manifest behaviors represent true bisexual arousal compared with sexual experimentation or curiosity? Considering that many male respondents in the Kinsey surveys were prison inmates or had jail experiences, what percentage represents a forced sexual activity or displacement activity? Pomeroy, a collaborator with Kinsey on both the 1948 and 1953 studies, gives as a rough estimate for males in their populations having jail or prison experience, 20% for those with an elementary school education, 15% for those with a high school education, and 10% of those with a college education (Pomeroy, personal communication, 1991). Gebhard (1972), another Kinsey coauthor, also discussed the lack of randomness and other shortcomings of the original volumes. He pointed out that the original research had a high percentage of individuals who had been incarcerated, particularly among the non-college-educated male population and this inflated the proportion of those with ambisexual experience in the total study population.

All the studies reported below, except those specifically of self-defined gays (see Table II), were random surveys of noninstitutionalized individuals.3 Pomeroy and Gebhard also remarked that the original Kinsey studies had many respondents associated with known gay groups. This too biased their sample. None of the randomized studies indicated they culled their samples of individuals of known gay organizations nor of those with histories of incarceration.

Last, it should be noted that the Kinsey studies used face-to-face interviews. Some of the following studies did similarly. Others, however, had respondents reply to the “sensitive” questions by completing a self-administered questionnaire, and others conducted their interviews by phone using random digit dialing (RDD) techniques to obtain their samples. One report (Hart, 1968) involved the investigator getting to know personally all members of a community.


United States of America

It is statistically randomized studies that should provide us the most reliable information (see Table I). Using data from a national area probability survey in 1970, bolstered by a National Opinion Research Center (NORC) survey in 1988, Fay et al. (1989) estimated that about one of five American males had ever-in-their-life any same-sex contact where one or the other reached climax. However, when they focused on those whose same-sex experiences were not rare or confined to adolescence but could be considered as at least occasional adult behaviors, they estimated homosexual activity a characteristic of only about 3.3% of the male population. They also conclude, after analyzing the data in several ways, the “nonrespondents, as a group, did not have a higher proportion with same-gender sexual experience than the rest of the sample.”

The National Opinion Research Center random household study of 1988 (Davis and Smith, 1988) found that only 2.4% of their males indicated same-sex behavior within the preceding year. And a 1991 NORC study (Smith, 1991) found less than 2% of sexually active adult males and females (about the same for both sexes in a 1988 survey and slightly more males in a 1989 survey) reported exclusive homosexual or bisexual activity the preceding year. After adjusting for “uncertain” or nonresponding cases Smith reported only 0.7% were exclusively homosexual since the age of 18 and an additional 5.6% bisexual.

A probability sample household survey in Dallas County, TX, conducted in 1989 (Dixon et al., 1991) for the Centers for Disease Control, concluded by combining data from a primary study and a follow-up on those originally not responding, that only 7.3% of their male population, ages 18-54, had any male-to-male sexual contact from 1978 to 1989; the years of interest due to the AIDS problem.

Using data from five NORC studies, the Dallas study, and a Research Triangle Institute study (1990), Rogers and Turner (1991), after an extensive analysis of the data, concluded that all the surveys were in substantial agreement and their findings point to a minimum of 5-7% of U.S. men involved in same-sex contact during adulthood.

A 1989 random sample of Hawai‘i State residents—Caucasians, Orientals, and Polynesians—interviewed by telephone using random digit dialing (RDD) techniques, found just about 3% of males and 1.2% of females as having engaged in same-sex or bisexual activity (Diamond et al.., in preparation). (The RDD technique used is described in Diamond and Dannemiller, 1989.) This is the only statewide survey known.

Most interestingly, several little known telephone surveys also using RDD, were reported by Schreiner (1986) for the National Organization of Gay and Lesbian Scientists and Technical Professionals, a Chicago-based organization (NOGLSTP). One study cited was commissioned by the San Francisco AIDS Foundation. On their behalf the Research and Decisions Corporation estimated that in 1984, 9.9% of San Franciscans identified themselves as homosexual or bisexual. Another RDD study Schreirier (1986) cited was one done by ABC News in 1985. The ABC research unit found that 9% of New York city respondents acknowledged gay or bisexual activity, whereas 4% of the national sample did so. The exact wording of the questions was not clear from the publication nor did it distinguish male from female.

Asia and Pacific

International studies provide figures within the same range. Four national studies of high school and University students from Japan, conducted in 1974, 1975, 1981, and 1987, found that between the age of 10 and 22, male same-sex contact, which may include bisexual contact, was reported by 7.1, 6.7. 5.8, and 3.7% of their male sample and 4.0, 3.8, 5.0, and 3.1% of their female respondents. The 1974 study was by Asayama (1976); the others were conducted by the Japanese Association for Sex Education (personal communication). All four studies used essentially the same techniques and questions.

From the Philippines are data from a full population rather than a survey. Hart (1968) did anthropological field work living for years in the village of Caticugan on the island of Negros, the province of Negros Oriental. During his stay he came to know all the village inhabitants and had many native informants. His work was analyzed by Whitam and Mathy (1986). In a village of 729 persons Hart found six male homosexuals and no lesbians. In the province of Siaton, which Hart also studied, with a population of 2862, there were 12 male homosexuals arid “several” lesbians (<2% male homosexuals). Of Dumaguete, the capital city of Negros Oriental, with a population of approximately 18, 000, Whitam and Mathy wrote: “there were 70 openly homosexual males and 58 lesbians” (0.7%).

Recent research from Thailand by Sittitrai et al. (1992) reported similarly. Same-sex contact was reported by 3.3% of the males interviewed but only 0.2% indicated such behavior was exclusively homosexual. Among females 1.1% reported they had same-sex contact and 0.9% said it was exclusively so. Whitam and Mathy reported (1986) on a nonrandom survey given to 178 male Thai students; 4.5% admitted to K =4-6 behavior.

New work from the Republic of Palau is also available. A 10% random sample of the total population, conducted by Morens and colleagues from the University of Hawai‘i and Polloi of the Palau Department of Health, as part of a general health survey, asked of sexual behavior for the preceding 12 months. They found that, after the age of 20, homosexual activity was reported by 1.9% of the males and 2.8% of the females interviewed. Bisexual activity was reported by 2.8% of the males and 0.7% of the females. The investigators noted that discussion of sexual behavior among the Palau is common and considered the responses reliable (Morens and Polloi, in preparation).4

From China (Liu, Ng and Chou, 1992) in the first wide scale survey of that nation, reports are that 7.6% of college students had homosexual experiences or attractions: 2.3% of “peasants” and 0.5% of “city dwellers” considered themselves gay. The authors themselves, however, consider these latter figures low.

Europe and Great Britain

Research cited by Schover and Jensen (1988), for a randomly selected population of Danish women personally interviewed by a female physician as part of a general preventive health study, found only 2 of 625 women from 22 to 70 years of age reporting having had a homosexual experience. (In comparison, if considering whether the women were hiding their behavior, more than 20% admitted to extramarital relations.)

Findings from a British national random study (Wellings et al., 1990) also show low figures for homosexual activity. These researchers report 9% of men and 4% of women reported ever having some homosexual experience, and only 5% of men and 1% of women reported ever having a homosexual partner.

A recent report by the Dutch National Institute for Social Sexological Research (N.I.S.S.Q.) and Utrecht University documented their national random sample of male and female sexual behavior (Zessen and Sandfort, 1991). In 1989, using face-to-face interviews, only about 12 in 100 males admitted to ever-in-their-life having a homosexual experience (“at least manual stimulation of the bare genitals of at least one partner”) and only about 8% said they had ever considered being homosexual. In the preceding 12 months only 3.6% of the males and 0.3% of the females surveyed had exclusive same-sex contacts and an additional 1.9% of their male sample and 0.5% of their female sample had bisexual activities.

Preliminary reports from France have indicated a survey done under the leadership of Alfred Spira from the Bicêtre Hospital found 4.1% of men and 2.6% of women said they had homosexual intercourse at least once in their lives. Only 1.1% of men and 0.3% of women said they had homosexual intercourse in the past 12 months (Aldhous, 1992).


Directing attention now specifically to studies of self-identified male homosexual populations we can gain additional information about bisexuality (see Table II). A study of several thousand self-identified homosexual males in the United States, the Netherlands, and Denmark by Weinberg and Williams (1974) reported about one in five American homosexuals were bisexually active as K = 2, 3, or 4 and about one in ten Dutch or Danish gay men were similarly active. Their findings for Americans, however, are at some variance with three more recent studies by Bell and Weinberg (1978), McWhirter and Mattison (1984), and Higa (1988). From these newer studies, of males who identify as engaging in same-sex behavior, about three or more out of four, claim to be exclusively homosexual and fewer than one in ten claim they ever had more than incidental sex with a female. Unfortunately, all four of these studies were nonrandom.



In 1972, Gebhard concluded “weighting by marital status we can say that about 4% of the white college-educated adult males are predominantly homosexual [K= 4-6) … the bulk of the 4% are exclusively homosexual.” That is certainly in the range of the figures found more recently. At that time he did not elaborate on data for the noncollege sample. In 1977, however, he wrote that if the original data set was culled of its figures for individuals from known gay organizations, it would still leave, for the general population, “13.95% of males and 4.25% of females” in the general population with extensive or at least more than incidental homosexual experiences (letter to National Gay Task Force quoted in Voeller, 1990). At this date, with our new data, such a range seems too large.

Differences between the findings of the original Kinsey et al. study (1948), the early Weinberg and Williams (1974) study, and the more recent research offered may certainly reflect behavior changes over time. The evolving so-called “Age of Aquarius” and movement toward decreased sexual restrictions might have fostered, among homosexually oriented males or females, a lessened feeling of obligation to engage in heterosexual activities, and for heterosexuals provided an increased availability of partners to more readily satisfy their erotic desires.

Another difference accounting for the variance from the Kinsey figures is that Kinsey et al. often used lifetime figures and these have been extrapolated as contemporary population figures. The former would obviously be larger than the latter.

The non-American studies, despite wide cultural differences, also report low ranges of homosexual activities: very different findings than would have been predicted by the original Kinsey studies. They are, however, in the range found by the recent city, county, state, and national surveys in the U.S. No study finds any figure reaching 10% for all respondents experiences with homosexual and bisexual activity combined. Considering the comparatively liberal attitudes toward homosexual or bisexual activity in Denmark, the Netherlands, Philippines, and Thailand, if not in Japan, Great Britain, or the United States, these data must be considered strongly. All these studies taken together indicate that bisexuality, and indeed homosexuality, are less common than previously considered. This appears whether the group surveyed are self-identified gays or from random samples.

It might also be argued, as done by Rogers and Turner (1991) that the figures reported are minimums and that most individuals who engage in homosexual and bisexual activities remain closeted for these surveys, preferring to retain their privacy despite pledges of confidentiality by researchers. This is certainly a possibility and might indeed hold for a certain percentage of respondents in those cultures where homosexuality is a subject of derision and even subject to prosecution as illegal. But evidence for this being widespread in the confidential surveys described above is not supported by the evidence of follow-up studies. Even Rogers and Turner’s own analysis of those who did not respond to the NORC sex behavior questions infers that the responses were probably reliable.5 They also admit “the response rates obtained in these surveys … were within or above the range typically obtained in such surveys.”

Nepon, the field director for the 1970 NORC study used by Fay et al. (1989) and Rogers and Turner (1991) noted that the refusal rate seemed no higher than normal and unrelated to the subject matter. It seemed more a function of the usual “too busy, ” “not interested, ” and “just leaving” (Fay et al., 1989). And with sophisticated analysis, using different weighting methods, the findings were similar from the 1988 and 1991 NORC studies and those used by Rogers and Turner (Davis and Smith, 1990; Smith, 1988). Fay et al. (1989) hypothesize that their data might, at most, be doubled to account for missing or falsely given data and still arrive at no more than a 6% figure for the percentage of adult males that might occasionally or regularly engage in same-sex activity as adults.

The Dallas researchers did a follow-up of the CDC-sponsored survey to check the reliability of their original findings. They did find some original reluctance by gay males (and intravenous drug users) to participate in the original survey, but found that once individuals agreed to participate there was no reason to believe they were not honest. Adjusting their data for projected nonresponse and nonresponse bias the CDC researchers felt raising their results to 7.3% from their original finding of 5.1% would account for any “closeting” (Dixon et al., 1991). It was reputed that during the Dallas study some gay community leaders advocated nonparticipation in the survey but how much effect this had is not known.6

The Dutch researchers, sensitive to these issues of nonresponse and report validity, also did a follow-up survey to evaluate the likelihood that respondents would fabricate or hide. Their follow-up of individuals that initially refused to be interviewed found males not to differ significantly from the main sample on any relevant variable (life-style, sexual risk behavior, sexual preference). And while the females that were not in the original study, but found in the follow-up, reported more AIDS risk behaviors and were more likely to have had homosexual experiences and a current homosexual self-labeling, the number involved was too small to estimate how much the basic figures might be modified (Zessen and Sandfort, 1991).7

Lastly we have to consider the findings from San Francisco and New York reported by the NOGLSTP (Schreiner, 1986). It might be hypothesized that the population of these cities, reputed to have the largest and most powerful gay, lesbian, and bisexual communities in the United States, would feel least inhibited in revealing same-sex activity and at least reach the 10% figure. This was not found.

In the village of Caticugan, where everyone was personally known to the investigator, closeting was unlikely. Hart (1968) reported there was some teasing of cross-dressers and effeminate males but homosexuals seemed open enough about their activities and the comments of others were more often playful than malicious. Hart concluded, “it is believed that the number of covert homosexuals is very small … the majority of residents of Siaton province and Caticugan are both lenient and indulgent of the local bayot [male homosexuals] and lakin-on [lesbians].” Others too have reported on the tolerance of the Filipino people toward homosexual behaviors (Guthrie and Jacobs, 1966). Since it is known that the Philippines of the 1980s are also relatively tolerant of homosexual behavior and homosexuality in general (Whitam and Mathy, 1986), there is little reason not to believe the findings. As Whitam and Mathy pointed out, even tripling the findings to account for covert homosexuality only brings the percentage to 2.1. These investigators, reviewing different cultures, consider 4-5% as the approximate range of homosexual activity to be expected in a population.

In any case, the oft-used 10% figure needs downward revision in light of the more reliable and valid data from the scientifically randomized samples reported here. It might be that Kinsey et al., with their more extensive and time-consuming face-to-face interviews, prompted more recall of early or incidental and long-forgotten activities. But this would not account for “missing” regular practices or patterns to reflect figures of active adult heterosexual, homosexual, or bisexual individuals. And certainly it must also be admitted that revealing homosexual activities today is socially and legally less threatening than it was 50 years ago.

An additional matter needs be considered to arrive at total population figures. Starting with the original Kinsey et al. studies, almost all research of male and female behaviors has found that the incidence of male homosexual activity greatly exceeds that of females, by a factor of approximately 2. In overall population statistics, then, any single figure should average the estimates rather than assume male and female homosexual expressions are equal in frequency.

Looking at the phenomenon from another perspective, it is safe to say that most individuals, most of the time, and for most of their lives, have sex with only males or females, not both. On the other hand, some individuals, at least for a few years, engage in sex with both and some minority engage exclusively in same-sex activities. To be cautious we also have to consider that a population of males and females—whose size we do not know—do not associate with gay groups or consider themselves as such, but do demonstrate homosexual or ambisexual activity. But considering the data now available, we might anticipate that the number is significantly smaller than previously thought.

To arrive at a usable “round” figure for those adults who regularly engage in, or have since adolescence at least once engaged in, same-sex activities we can consider together all the non-Kinsey data reported above. If we use the largest figures available from statistical weighting for nonresponse or response ambiguity, and allow for study differences, giving equal weight to all reports as interpreted within each study, we find a mean of 5.5% and a median of 5.3% represents the population of males having engaged in same-sex behavior.8 Female figures are even smaller: mean 2.5% and median 3.0%. Our round numbers would thus be 5-6% for males and 2-3% for females. In modeling AIDS epidemiological transmission or risk factors and for other population considerations, these might be the range of numbers most applicable.


Kinsey et al. (1948) reported, in one of their most controversial findings: “it appears that nearly half (46%) of the [male] population engages in both heterosexual and homosexual activities, or reacts to persons of both sexes, in the course of their adult lives” (p. 656). By extension, with heterosexuals (K = 0) representing 50% of their population and exclusive homosexuals (K = 6) 4% of their population, ambisexuality could be considered almost as common as heterosexuality. Figures given by Kinsey et al. (1953, Table 142) for female ambisexual behavior, K = 2-5, were something between 4 and 11% for those 20 to 35 years of age. Some 1-3% demonstrated exclusively homosexual, K = 6, activities. The validity of these numbers have been challenged before (Cochran et al., 1964; Reisman and Eichel, 1990; Terman, 1948; Wallis, 1948). Unfortunately, studies of female sexual activity are rarer than those of males. From the studies above, few of which asked of it specifically or in detail, it is difficult to estimate adequately the size of the bisexual population. Certainly it would be a portion of the total group that engaged in same-sex behaviors. And again, in modeling AIDS epidemiological transmission or risk factors and for other population considerations, these might be the range of numbers most applicable.

Additional Comments

Many of the studies above found a significant portion of their respondents had not had any sexual activity during the preceding 12 months. Also, an oft-neglected finding of Kinsey et al. (1948, 1953) is that a percentage of male and female respondents were not sexually aroused by individuals of either sex. Such individuals were recorded as X, off the Kinsey scale, for sexual orientation. These persons may abstain from sexual activity or, on the other hand, engage in sexual activities for nonerotic reasons. Among females this may represent as much as 15% of the population. Most more recent studies, unfortunately, are silent on X-identified individuals so these Kinsey figures can not be updated. Certainly then, this leaves them much more flexible in why and with whom they might engage in sexual activities. For this population nonsexual motives may become more important than any erotic drive. These nonsexual motives may fuel a significant amount of female bisexual activity. Elsewhere Diamond (1993) reviewed different findings indicating that bisexuality, compared with exclusive heterosexuality or homosexuality, appears more susceptible to social forces. This too might, considering the fear of AIDS, reduce the incidence of non-erotic-motivated bisexual activity. On the other hand, the comparatively high incidence of HIV infection among homosexual and bisexual men. some 57% of all those persons in the U.S. with AIDS (Centers for Disease Control, 1993), may foster sex with females while a male partner might be preferred.

Another issue which is relevant to the present overall discussion is that studies over the last several years are calling attention to the genetic component to homosexual orientation (e.g., Bailey and Pillard, 1991; Whitam, Diamond, and Martin, 1993).This intimates that these population figures may be fairly constant and not too susceptible to social changes.

This discussion of population percentages and sexual activities, while of scientific import, must also be recognized as having policy and political implications. These lower numbers alone can account for the downward revised estimates of the growth curves for the AIDS epidemic. They may also help plot the future course of the disease. With politics, it may reflect the type of gay aligned or opposed support AIDS research and treatment may develop or how the interests of homosexuals or bisexuals are considered. In all cases, it is to be hoped that the science should overcome the politics.9


A shared appreciation is sent to those colleagues who shared their prepublication data and thanks to those reviewers who helped make this a better paper.



1 University of Hawai‘i, John A. Burns School of Medicine, Department of Anatomy and Reproductive Biology, 1951 East-West Road, Honolulu, Hawai‘i 96822.

2 Kinsey reserved the term bisexual for biological features or behaviors. He preferred the term ambisexual for one who might eroticize males or females.

3 The Japanese studies were, however, limited to postpubertal student populations. They are nevertheless still felt useful since they are of repeated measures to different populations over time using essentially the same instrument. They arc also the only known surveys available for that country. The studies of gay populations, by their nature, essentially used opportunity/convenience or snowball samples.

4 In these nonhomophobic societies, Philippines. Thailand and Palau, it is interesting to note we find reported the lowest rates of same-sex activity. And while the number of Thai males having same-sex contacts is almost three times the number of females reporting similarly, this and the Palau research are the only known studies which report exclusive homosexuality in a larger proportion of females.

5 This is not to argue that this is not a valid consideration. Sex behavior research is increasingly under scrutiny for methodological improvement to increase response accuracy (Catania et al., 1990).

6 I have been involved with AIDS-related research and community service since the early 1980s. At that time also many gay leaders advocated noncooperation with researchers. However, their admonitions seemed to have little influence on our work. Now, indeed, the gay community is quite cooperative.

7 There is always a question of honesty or candor in surveys or interviews regarding highly private material. In my experience with research of illegal or unpopular activity such as pre- and post-legal abortion (Diamond et al., 1973; Steinhoff and Diamond, 1977) it was found, with external verification, that there was no reason to believe those who refused our interviews to be any different than those who responded. Similarly with surveys of attitudes toward pornographic material (Diamond and Dannemiller, 1989). I feel reliable responses are more probably a factor of the interviewer skills, believability, and persuasiveness than the respondents’ hesitation.

8 Since several of the studies reviewed analyzed the same NORC data sets it is accepted that these figures are thereby skewed. The NORC surveys did not report on women.

9 While this paper was in press Bailey et al.. (1993) reported on a survey of 3, 224 men aged 20-39 in the U.S. conducted by Battelle Human Affairs Research Center. They report 2.3% of sexually active men had any homosexual activity during the years 1981-1991 with only 1.1% reporting exclusive same sex activity. These low figures attracted a great deal of attention and prompted a public critique by the polling firm of Louis Harris & Associates (Taylor, 1993). Two methodological features of the Battelle poll seem particularly implicated in the low findings: first, Battelle used only female interviewers who might have inhibited full disclosure by the male respondents and second, they interviewed those of a narrow age range.


Aldhous, P. (1992). French venture where U.S. fears to tread. Science 257: 25.

Asayama, S. (1976). Sexual behavior in Japanese students: Comparisons for 1974, 1960, and 1952. Arch. Sex Behav. 5:371-390.

Bailey, J. M., and Pillard, R. C. (1991). A genetic study of male sexual orientation. Arch. Gen. Psychiat. 48 (December): 1089-1096.

Bell, A. P., and Weinberg, M. S. (1978). Homosexualities—A Study of Diversity Among Men and Women. Simon and Schuster, New York.

Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981). Sexual Preference—Its Development in Men and Women. Alfred C. Kinsey Institute of Sex Research. Bloomington, IN.

Billy, J. O. G., Tanfer, K., Grady, W. R., and Klepinger, D. H. (1993). The sexual behavior of men in the United States. Fam. Plan. Perspectives. 25: 52-60.

Booth, W. (1929). U.S. probe meets resistance. Science 244: 419.

Buchanan, P. J.(1990). Sex, lies and Dr. Kinsey. San Francisco Examiner, October 23.

Catania, J. A., Gibson, D. R., Chitwood, D. D., and Coates, T. J.(1990). Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol. Bull. 102: 339-362.

Centers for Disease Control (1993). HIV/AIDS Surveillance Report. February 9.

Cochran, W. G., Mosteller, F., and Tukey, J. W. (1964). Statistical Problems of the Kinsey Report on Sexual Behavior in the Human Male, American Statistical Association. Washington, DC.

Coolfont Report. (1986). A PHS plan for prevention and control of AIDS and the AIDS virus. Public Health Rep. 101: 341-348.

Davis, J. A., and Smith, T. W. (1988). NORC’s 1988 General Social Survey. General Social Surveys, 1972-1988: Cumulative Code Book. National Opinion Research Center, Chicago.

Davis, J. A., and Smith, T. W. (1990). General Social Surveys. 1972-1990: Cumulative Code Book, National Opinion Research Center. Chicago.

Deisher, R. W. (1989). Adolescent homosexuality: Preface. In Herdt, G. (ed.). Gay and Lesbian Youth, Haworth. New York. pp. xiii-xv.

Diamond, M. (1992). Sexwatching: Looking at the World of Sexual Behavior, London, Prion Press Ltd.

Diamond, M. (1993). In E. Haeberle (ed.), Bisexuality: Biological Aspects. Bisexualities. Walter de Gruyter, Berlin, in press.

Diamond, M., and Dannemiller, J. E. (1989). Pornography and community standards in Hawai‘i: Comparison with other states. Arch. Sex. Behav. 18: 475-495.

Diamond, M., Ohye, R., and Wells, J. (In preparation). Population survey of Hawai‘i: Relation to AIDS. AIDS: Educ. & Prevention.

Diamond, M., Steinhoff, P. G., Palmore, J. A., and Smith. R. G. (1973). Sexuality, birth control and abortion. J. Biosoc. Sci. 5: 347-361.

Dietz, K. (1988). Statistical Analysis and Mathematical Modeling of AIDS, Oxford Medical Publications, Oxford, U.K.

Dixon, B. W., Streiff, E. J., and Brunwasser, A. H. (1991). Pilot study of a household survey to determine HIV seroprevalence. Morbidity and Mortality Weekly Report 40: 1-5.

Fay, R. E., Turner, C. F., Kiassen, A. D., and Gagnon, J. H. (1989). Prevalence and patterns of same-gender sexual contact among men. Science 243: 343.348.

Gargon, E. A. (1991). Coming out in India, with a nod from the gods. New York Times (International ed.), August 15.

Gebhard, P. H. (1972). Incidence of overt homosexuality in the United States and Western Europe. NIMH Task Force on Homosexuality: Final Report and Background Papers, National Institute of Mental Health. Rockville, MD. pp. 22-29.

Green, R. (1987). The “Sissy Boy Syndrome” and the Development of Homosexuality, Yale University Press, New Haven. CT.

Guthrie, G. M., and Jacobs, P. J. (1966). Child Rearing and Personality Development in the Philippines. Pennsylvania State University Press, University Park.

Haeberle, E. J. (1978). The Sex Atlas: A New Illustrated Guide. Seabury Press, New York.

Harris Poll (1988). Survey for Project Hope. Louis Harris & Associates, New York.

Hart, D. V. (1968). Homosexuality and transvestism in the Philippines. Behav. Sci. Notes 3: 211-248.

Higa, D. (1988). The psychosocial functioning of individuals at risk for AIDS. Master’s thesis, University of Hawai‘i, School of Social Work.

Kinsey, A., Pomeroy, W., and Martin, C. (1948). Sexual Behavior in the Human Male.W. B. Saunders, Philadelphia.

Kinsey, A., Pomeroy, W., Martin, C., and Gebhard, P. (1953). Sexual Behavior in the Human Female. W. B. Saunders, Philadelphia.

Liu, D. L., Ng, M. L., and Chou, L. P. (1992). Sexual Behaviour in Modern China: A report on the nation-wide “sex civilisation” survey on 20, 000 subjects in China. Shanghai, San Lian Bookstore Publishers.

Marshall, R. (1991). A blow below the belt. Newsweek, July 1, p. 39.

McWhirter, D. P., and Mattison, A. M. (1984). The Male Couple: How Relationships Develop. Prentice-Hall, Englewood Cliffs, NJ.

Michael, R. T., Laumann, E. O., Gagnon, J. H., and Smith, T. W. (1988). Number of sex partners and potential risk of sexual exposure to human immunodeficiency virus. Morbidity and Mortality Weekly Report 37: 565-567

Morens, D. M., and Polloi, H. (in preparation). Sexual experience and health risk in the Republic of Palau.

Reisman, J., and Eichel, E. W (1990). Kinsey, Sex and Fraud: The Indoctrination of a People. Lochinvar-Huntington House. Lafayette, LA.

Research Triangle Institute (1990). National Household Seroprevalence Survey: Feasibility Study Final Report. Research Triangle Park, NC.

Rogers, S. M., and Turner, C. F. (1991). Male-male sexual contact in the U.S.A.: Findings from five sample surveys, 1970-1990. J. Sex. Res. 28: 491-519.

Schover, L. R., and Jensen, S. B. (1988). Sexuality and Chronic Illness: A Comprehensive Approach. Guilford, New York.

Schreiner, J. (1986). Measuring the Gay and Lesbian Population. National Organization of Gay and Lesbian Scientists and Technical Professionals. Report P.O. Box 14138. Chicago.

Schwager, S. J., Castillo-Chavez, C., and Hethcote, H. (1989). Mathematical statistical and mathematical approaches in HIV/AIDS modeling: A Review. In Castillo-Chavez. C. (ed.) Statistical approaches to AIDS Epidemiology. Springer-Verlag, Berlin.

Sittitrai, W., Brown, T., and Viruirak, S. (1992). Patterns of bisexuality in Thailand. Bisexuality and HIV/AIDS. In Tielman, R., Carballo, M., and Hendricks, A. (eds.). Prometheus Books, Buffalo, pp. 97-117.

Smith, T. W. (1988). A methodological review of the sexual behavior questions in the [NORC]. 1988 General Social Survey: GSS Methodological Report No. 58. NORC, Chicago.

Smith, T. W. (1991). Adult sexual behavior in 1989: Number of Partners, Frequency of intercourse and risk of AIDS. Fam. Plan. Perspect. 2:102-107.

Steinhoff, P. G., and Diamond, M. (1977). Abortion Politics. University of Hawai‘I Press, Honolulu.

Taylor, H. (1993). Number of gay men more than 4 times higher than the 1 percent reported in a recent survey. The Harris Poll #20: 1-4.

Terman, L. M. (1948). Sexual behavior in the human male; Some comments and criticisms. Psychol. Bull. 45: 443-59.

Voeller, B. (1990). Some uses and abuses of the Kinsey scale. In McWhirter, D. P., Saunders, S. A., and Reinisch, J. M. (eds.). Heterosexuality/Homosexuality: Concepts of Sexual Orientation. Oxford University Press, New York. pp. 32-38

Wallis, W. A. (1948). Statistics of the Kinsey Report, J. Am. Stat. Assoc. 44: 463-484.

Weinberg, M., and Williams, C. (1974). Male HomosexualsTheir Problems and Adaptations. Oxford University Press, New York.

Wellings, K., Field, I., Wadsworth, A. M., Johnson, A. M., Anderson, R. M., and Bradshaw, S. A. (1990). Sexual lifestyles under scrutiny. Nature 348: 276-278.

Whitam, F. L., Diamond, M., and Martin, J. (1993). Homosexual orientation in twins: A report on 61 pairs and three triplet sets. Arch. Sex. Behav. 22: 187-206.

Whitam, F. L., and Mathy, R. M. (1986). Male Homosexuality in Four Societies: Brazil, Guatemala, the Philippines and the United States. Praeger. New York.

Yamaguchi, M. (1991). Gay Japanese Cautiously Come Out of Closet. New York Native 6.

Zessen, G. van, and Sandfort, T. (1991). Seksualiteit in Nederland (Sex in the Netherlands). Swets & Zeitlinger, Amsterdam, The Netherlands.

Back to top