Well: How Many Americans Are Lesbian, Gay or Bisexual?

As it has for decades, the Centers for Disease Control and Prevention last week released its annual National Health Interview Survey on the health of Americans. But this year, there was a difference: For the first time, the respondents were asked about their sexual orientation.

Of 34,557 adults ages 18 and older, the survey reported, 1.6 percent said they were gay or lesbian. Some critics say the numbers are low, but they fall in the range of other surveys. In the new survey, however, only 0.7 percent of respondents described themselves as bisexual; other studies have reported higher numbers.

Adults who identified themselves as gay, lesbian or bisexual reported some different behaviors and concerns — for example, more alcohol consumption and cigarette smoking — than those who said they were straight.

But it can be difficult to elicit information that many people consider private. The New York Times spoke about such challenges with Gary J. Gates, a demographer at the Williams Institute at the U.C.L.A. School of Law, which focuses on law and policy issues related to sexual orientation and gender identity. Some of Dr. Gates's findings were echoed in the new survey. This interview was edited and condensed.

How was this survey conducted?

Survey takers had a computer that guided them through questions which they asked the respondent in person, and they used flash cards to show them potential answers.

Why do you think the figure for bisexuality was lower than in other surveys?

There is evidence that bisexuals perceive more stigma and discrimination than gay and lesbian people. They are much less likely to tell important people around them that they are bisexual.

The way this survey was conducted, the stigma experienced by bisexuals may have mattered more in their reluctance to tell the survey takers than it did for gay and lesbian people. If it were an anonymous interaction by phone or Internet, the stigma might not be as important. And N.H.I.S. did report that bisexuals reported high levels of psychological distress.

Were there other weaknesses in the survey?

They measured sexual orientation: straight, gay, lesbian or bisexual. But they didn't measure gender identity. They didn't ask, "Are you transgender?" or "Do you present your gender differently than the sex that was recorded on your birth certificate?" That is a serious gap.

You have written about best practices for surveying L.G.B.T. People. What are some considerations?

It depends on what question you want answered. When we ask about sexual orientation, we say, "Do you think of yourself as gay, straight or bisexual?" But you might want to focus on sexual behavior and health risk. Then you get a very different group than when you measure people's sexual identity.

In some surveys, two-thirds of those who say that they have had some same-sex sexual encounters would identify as heterosexual. I don't take a stand on whether they are, just that they reported same-sex behavior.

So how people identify themselves to survey takers reflects just that?

Yes. In the N.H.I.S., a little over 2 percent identify as lesbian, gay or bisexual. That's different than saying a little over 2 percent of the population is lesbian, gay or bisexual.

Some critics say that such efforts do not capture minority communities, where there may be resistance to complying with government survey takers.

There isn't much difference in the racial and ethnic characteristics of the straight population and the L.G.B. population in the N.H.I.S. There is a popular conception that there is a lower willingness to report L.G.B. identities among minorities, but very few surveys find that to be the case.

I don't want my words to ever be construed that I don't believe there is some portion of the population who is unwilling to disclose their identities to survey takers. But that's a largely unknowable number, because you're asking how many didn't do something.

Conversely, are there some groups who will disclose more readily than others?

In the Gallup data I've worked with, if you're under 30, 6 percent identify as L.G.B.T. If you're older, it's closer to 3 percent. But it's not that proportionally there are more L.G.B.T. people among the young, but that young people are more likely to disclose.

What about the 1.1 percent that the N.H.I.S. Said responded, "I don't know the answer" or something else?

The N.H.I.S. followed up with them. Thirty percent said they hadn't figured out their sexual orientation. Another 30 percent said they didn't understand the words gay, straight or lesbian. And 30 percent just refused to answer the question or reiterated, "Don't know."

How can survey takers obtain more reliable numbers?

We get accurate responses from people who are willing to identify. Is that accurate in terms of this almost existential thing we call sexual orientation? No. There are people who internally think of themselves that way who are not willing to tell the survey taker.

But some surveys then ask, "Who have you told about your sexual orientation?" And a small number respond, "No one." So there are people who will just identify themselves to a survey taker but to no one else in their life.

My argument is that these surveys are accurate portraits of who is willing to disclose in that setting, but it's not an accurate assessment of, for lack of a better word, the closet.