The Myth of Gay & Lesbian Affluence.
A recent ad run by the National Gay Newspaper Guild touts "Gay Market Muscle" and shows a well-dressed young white man flexing his biceps to represent the high average incomes of the Guild's readers. And while this group's readers may indeed be more affluent than gay men in general, the ad is also evoking an image of gay male affluence that has become widely accepted in the world of gay and lesbian marketing.
Over the last ten years, marketing companies have commissioned a number of studies on gay and lesbian consumers to aid the sale of marketing services such as list rentals and advertising space in gay publications. Two research firms in particular, the Simmons Market Research Bureau and Overlooked Opinions, have conducted surveys that are often cited to show that gay men and lesbians have higher incomes than the average American. A closer look at the work of those firms shows that such general conclusions about the gay community are not warranted by the data.
The Simmons survey is cited most often, thanks to a 1991 article about the survey in The Wall Street Journal (July 18, 1991). A few years earlier, Rivendell Marketing, a company that sells advertising space in gay and lesbian publications, had commissioned Simmons to survey readers of gay magazines and newspapers.
The men and women who responded to the first surveys in 1988 and 1989 had an average annual personal income of $36,000 and a household income of $55,430, far higher than the American mean. A 1992 survey of readers from the six papers in the National Gay Newspaper Guild showed even higher incomes: $41,300 for individuals and $63,700 for households. A more recent Simmons study, conducted in 1996, found that 28 percent of gay people surveyed had personal incomes of over $50,000, and 21 percent were in households with over $100,000 in income ("Affluence of Gay Market Confirmed," PR Newswire, Feb. 3, 1997). The latter study, which was conducted with the Mulryan/Nash advertising agency, drew respondents from the mailing lists of a political organization, a mail order catalog, and a credit card company.
Overlooked Opinions, a now defunct marketing organization, compiled a list of lesbian and gay people from sign-up sheets at events, such as the 1993 March on Washington for Lesbian and Gay Rights, and from other sources, such as bars or bookstores. Periodically they conducted a survey of a sample of people on their list, and those individuals were also found to have higher-than-average incomes. In a 1991 survey, gay men had an average income of $42,689 and lesbians of $36,072.
These findings are still taken widely for granted today. For companies attempting to identify new target markets--to the extent that they're seeking ways to reach relatively affluent people, such as readers of magazines--these surveys remain perfectly adequate. Similarly, companies marketing to more affluent gay people can target them using these mailing lists. But almost any other use of these data would be misleading. Neither organization conducts its surveys in a way that allows its findings to be generalized to all gay and lesbian Americans. We know from other research that readers of newspapers and magazines are significantly better educated and more affluent than the average adult in the U.S. For instance, male readers of USA Today earned $29,428 in 1989 and male readers of The Wall Street Journal earned $42,040, but both groups earned much more than the average individual male income of $19,893. So it's not surprising that the gay men and lesbians surveyed from samples of magazine readers have higher-t han-average incomes compared to all Americans, but this does not translate into higher average incomes for gay and lesbian people in general. By the same token, a sample of people obtained from a march on Washington is anything but representative of the whole gay and lesbian population, since not everyone can afford to attend such an event.
Toward a Truer Picture
Needed for a more accurate picture of gay and lesbian affluence would be a survey whose respondents were drawn from a more truly representative sample of the GLBT population in the U.S. Unfortunately, none of the large-scale population surveys conducted by federal agencies, such as the Census of Population or the Current Population Survey, directly asks respondents about their sexual orientation. However, the Census allows for some self-identification, while some random samples of the U.S. population have at times asked questions that allow us to identify at least some parts of the GLBT population. This report is based on data collected from seven studies. (A brief synopsis of each is provided in the sidebar.) Three of these studies collected information on individual incomes: the General Social Survey (GSS), the 1990 Census, and the Yankelovich Monitor.
A pattern emerges across the three surveys for both men and women. The average lesbian or gay man earns no more than the average heterosexual woman or man, and in some cases, gay people earn less on average. Figures 1 and 2 summarize the averages from the two studies. The average gay man earns from 4 percent to 7 percent less than the average heterosexual man. The average lesbian earns less in the GSS and roughly the same income as heterosexual women in the Yankelovich Monitor. However, according to the Census, the average woman with a female partner earns more than the average heterosexual woman does. Looking more closely, that apparent income advantage disappears once we take into account the fact that lesbians work more hours per week and more weeks per year than the average heterosexual woman.
A more sophisticated comparison of gay, lesbian, and bisexual individuals' incomes with heterosexuals' incomes involves taking other important factors into account, such as education, geographical location, occupation, experience, and race. Two recent studies have applied this statistical procedure (known as multiple regression) to the General Social Survey and the 1990 Census. M. V. Lee Badgett's 1995 analysis of the GSS found that gay men who worked full-time earned as much as 27 percent less than comparable heterosexual men. Lesbians who worked full-time earned roughly the same as heterosexual women. Marieka Klawitter and Victor Flatt's study (1998) of the Census data found very similar results. Men with male partners earned 26 percent less than did married men with the same education, location, race, age, number of children, and disability status. But, as in the GSS, women with female partners showed no difference in earnings compared to heterosexual women once the other factors are taken into account. Th e findings for gay men strongly suggest the influence of workplace discrimination.
The Special Case of "Household" Income Even though gay men might face an earnings disadvantage as individuals and lesbians might experience no consistent earnings advantage, it might be argued that two gay men or two lesbians could still have more discretionary income than a heterosexual couple. The stereotype of gay or lesbian couples as "DINKs" ("dual-income-no-kids") involves two assumptions: first, that same-sex couples will have higher incomes than married couples; and second, that lesbians and gay men have no children. The first assumption is thoroughly contradicted by the data.
Five of the scientifically sound surveys of lesbian, gay, bisexual, and heterosexual people offer information on family or household incomes. Figure 3 shows the 1990 Census and Yankelovich results. Some of the surveys, represented in Figure 4, asked respondents to indicate which of several income categories their household income fell into, so averages cannot be compared directly a man's income, so we would expect the average household income of a female couple to be less than the income of a typical male-female couple, while two average men sharing a household would have an even higher income because of this gender effect. Klawitter and Flatt (1998) further analyzed the differences in household income, controlling for the couple's geographical location, education, age, race, sex, and other relevant variables. Once these factors were taken into account, married couple households and male same-sex couple households had roughly equal household incomes, while female same-sex couples brought home 18 to 20 percent less income than a similar married couple.
The remaining four survey sources tell much the same story. In each case, lesbian, gay, and bisexual people are spread throughout the range of household income distribution for all householders. Far from being concentrated at the high income end, a significant number of gay people are clustered at the lower end of the distribution. (However, this could be the result of different household sizes for heterosexual and gay households.)
Finally, as with myths on income, the myth that gay and lesbian people do not have children to care for, making them those idealized DINKs that marketers love, is called into doubt by two of the surveys discussed here. In the 1992 Voter News Service (VNS) voter exit poll, 31 percent of lesbians compared to 37 percent of heterosexual women had children under 18 living at home, as did 23 percent of gay men and 33 percent of heterosexual men. In the CARDIA (1992-93) Lake Research Inc. / HRC Yankelovich Monitor, 67 percent of lesbians and 72 percent of heterosexual women were parents; 32 percent of lesbians and 36 percent of heterosexual women had children under 18 in the household. Roughly half as many gay as heterosexual men were either parents (27 percent of gay men and 60 percent of heterosexual men) or had children under 18 in their households (15 percent of gay men and 28 percent of heterosexual men), however. The surveys do not allow us to tell whether the children in gay or lesbian respondents' households were from prior heterosexual relationships, were adopted or co-parented by the respondents, or were conceived through alternative insemination. Regardless, the presence of children is clearly common in lesbian and gay households, suggesting that lesbians and gay men have economic responsibilities for children.
How to Count Americans Who Aren't Out?
Gay Americans have long been a hard to measure population, hidden as they often are by fear, discrimination, and secrecy. It is not, therefore, surprising that one concern likely to emerge about these surveys is their accuracy in identifying lesbian, gay, and bisexual people. The measurement and survey issues are certainly tricky, since one might expect that many people would be reluctant to identify themselves as gay to a stranger on the phone, at their front door, or outside a voting booth. For the purposes of income comparisons, a failure to answer accurately will not present a problem unless levels of disclosure are related to levels of income. For instance, if people with lower incomes are less likely to report that they are gay, then the average income for this group reporting their sexual orientation will be higher than it is for this group as a whole. If, instead, higher income gay people do not disclose their sexual orientation, then the average we see on a survey will be lower than that of the true gay population.
A second issue is that different people will define what it means to be "gay" or "lesbian" or "bisexual" differently. For some, being gay means having sexual partners of the same sex, while others might say that they are gay because they live and socialize with mostly gay people and are active in life in the gay community. Other possible definitions exist, as well.
Fortunately, these two issues do not pose a serious problem in the conclusions drawn in this paper about the economic position of lesbian, gay, and bisexual people. The surveys reviewed here use a variety of definitions for lesbian, gay, or bisexual people including self-identification, sexual behavior, and social behavior (having a same-sex unmarried partner) and all reach similar conclusions. Further, all but one of the surveys reviewed here were designed to allow respondents to indicate their sexual orientation without having to come out directly to an interviewer. These two factors should minimize concerns about survey confidentiality and definitions.
We also do not know whether or how income affects gay people's willingness to disclose their sexual orientation. Individuals with higher income might have more to lose if they experience discrimination as a result of disclosure, but they may also have the resources to avoid serious harm. If gay people with lower incomes are less willing to reveal their sexual orientation, this would artificially boost the average incomeof those who are sampled.
For all these caveats, one is struck by the robustness of the central finding, the consistency of results from several studies despite their differing definitions and sampling methods. And what they show is that, on balance, GLBT people do not earn more as individuals than heterosexual people do, and they do not live in more affluent households. Two studies suggest that gay men earn less than similarly qualified heterosexual men. But what all these studies show is that GLBT people are distributed throughout the spectrum of income distribution for all Americans. Given this consistency, it's puzzling that people continue to rely on misleading marketing statistics, especially when the methods are now available to produce a broadly reliable picture of gay and lesbian affluence relative to the American populace as a whole.
M. V. Lee Badgett teaches economics at the University of Massachusetts at Amherst. She is also the Director of the Institute for Gay and Lesbian Strategic Studies.
Klawitter, Marieka M. and Victor Flatt. "The Effects of State and Local Antidiscrimination Policies for Sexual Orientation," Journal of Policy Analysis and Management, Vol. 17, No.4, Fall 1998.
Badgett, M.V. Lee. "The Wage Effects of Sexual Orientation discrimination," Industrial and Labor Relations Review, Vol. 48, No. 4, July 1995.
Badgett, M.V. Lee. "Vulnerability in the Workplace: Evidence of Anti-Gay Discrimination," Angles: The Policy Journal of the Institute for Gay and Lesbian Strategic Studies, Vol. 2, No. 1, September 1997.
Sources of Survey Data Cited in Charts
1990 U.S. Census of Population
The Census is a self-administered questionnaire sent every ten years to each household in the United States. A self-designated "householder" is the reference person for all other household members. In 1990, the Census allowed a household member to be designated as an "unmarried partner" of the householder, so people with same-sex partners are identifiable. However, single lesbians and gay men and those who do not live with their partners are still not identifiable. The questionnaires are not anonymous, but information is recoded and location variables are highly aggregated so that the Census Bureau can provide publicly available random samples that ensure the confidentiality of the individual identities. Researchers Marieka Klawitter and Victor Flatt's analysis of the 1990 data was used in this report. Only 0.18 percent of households contained same-sex unmarried partners, and Klawitter and Flatt used a sample of roughly 13,000 married couples and 6,800 same-sex couples. The income reported was for 1989.
The General Social Survey
The General Social Survey (GSS) is an annual survey conducted by the National Opinion Research Center (NORC) that, in 1989, began including questions on the sex of respondents' sexual partners since their eighteenth birthday. The questions on sexual behavior are asked on a self-administered questionnaire that is returned to the interviewer in a sealed envelope. Sexual orientation can be measured by identifying people with same-sex sexual partners. The author of this paper pooled the data from 1989 to 1991 and analyzed the 1,680 full-time workers in that subsample. Of this group, 4.8 percent reported having had at least one same-sex sexual partner, and 3.0 percent reported at least as many same-sex as opposite sex partners.
This is an annual, face-to-face survey of attitudes and demographics. In 1993, Yankelovich Partners included a new option in a list of descriptions under the heading, "Which of the following words or phrases describe you?" Respondents were given a card with a list of such words and phrases. The thirty-third possibility was "gay / homosexual / lesbian," which was preceded by "a pleasure seeker" and "tense" and followed by "religious" and "witty." Respondents were asked to call out the numbers corresponding to the descriptors, and the interviewers recorded the numbers. The total sample in 1993 was 2,503 respondents, including 65 gay men (5.2 percent of all men) and 83 lesbians (6.6 percent of all women). Marketing analyst Grant Lukenbill reported the summary statistics from a report prepared by Yankelovich Partners.
Lake Research / Human Rights Campaign
ICR Survey Research Group conducted a series of polls in 1996 for Lake Research, Inc. and the Human Rights Campaign, a national gay political organization. Using random-digit dialing, ICR interviewed 28,282 individuals who were asked, among other things, "Are you yourself (for males) gay or bisexual / (for females) a lesbian or bisexual?" Of those surveyed over three months, 354 individuals identified themselves as gay, bisexual, or lesbian (1.3 percent of those interviewed). Gay respondents were asked about their attitudes and opinions on several issues, and basic demographic data, including pretax household income, were collected.
Coronary Artery Risk Development in Young Adults (CARDIA)
The CARDIA study is a longitudinal epidemiological study of young black and white men and women conducted since 1985. Respondents were chosen randomly in Chicago, Minneapolis, and Birmingham (Alabama), and from the membership of a large health plan in Oakland, California. In 1989 researchers sent participants a self-administered questionnaire on HIV-related issues, and included questions on lifetime numbers of sex partners, both same-sex and opposite-sex. The survey data were matched with data from respondents' health exams by public health researchers Nancy Krieger and Stephen Sidney between 1992 and 1993. Overall, 12 percent of the 1,724 respondents (11 percent of women and 13 percent of men) reported at least one same-sex sexual partner.
Voter News Service Exit Poll
The Voter News Service data were collected from a sample of at least 300 different voting places on election day in 1996. Trained interviewers distributed surveys to a random sample of individuals leaving the polls. The questionnaire was self administered, and respondents returned the surveys to a box. In 1992, the questions asked included, "Do Any of the Following Apply to You?" followed by a list of characteristics, including "gay / lesbian / bisexual." Respondents could check a box for either "yes" or "no," but non-responses were coded as "no." In 1996, the question was asked separately: "Are you) Gay, Lesbian, or Bisexual?: Yes or no." Although voters tend to have higher levels of education and income than non-voters, that effect should be similar for gay and heterosexual voters, making comparisons useful.
The 1992 income data reported here are based on an analysis of the data by Murray Edelman of Voter Research and Surveys, and the 1996 data come from unpublished tabulations by Rutgers University political scientist Robert W. Bailey. In 1992 the combined national and California and New York surveys yielded a sample of 260 gay men (3.3 percent of all male voters) and 206 lesbians (2.3 percent of female voters), In 1996, the sexual orientation question appeared on fewer surveys and only 196 individuals identified themselves as gay, lesbian, or bisexual.