Your question was: ‘are LGBT people less likely to respond to report their sexual orientation in surveys than heterosexual people’. The first thing to acknowledge is that LGBT identities are/ have been stigmatised and privatised. In the UK the Sexual Offences Act 1967 decriminalised homosexuality, as long as it was conducted in private between two consent adults. This legislation thus linked identities with behaviours and enforced a privatised identity for LGBT people. Similar legislation prohibiting or limiting LGBT behaviours/ identities exists or has existed in other countries (see currently in Uganda and India but also Italy and many US states). This has never occurred for heterosexual identities, as far as I am aware, anywhere in the world and so there has never been a reason not to disclose a heterosexual sexual orientation.

In the UK, when the Census first asked a question about ‘race’ and ethnicity there was a similar reluctance to disclose Black and Minority ethnic identities (but not white to the same degree) to a government sponsored survey. Now many people believe that the UK census gives some reliable estimates as to their inclusion in the UK population.

There are relatively few government sponsored surveys in the UK which ask questions about sexual orientation or gender identity. I successfully lobbied the Department of health to include LGB in a large scale health survey but there were many ‘refuse to answer’ the question. This may be partly attributed to the fact that people are not used to being asked the question and partly they have not been assured to the uses of the data. Because we do not know the size of the LGBT population we do not know whether these are more likely to be LGBT people or not, but in the light of the above discussion one might expect that this would be more likely.

An interesting survey is the NATSAL published in the Lancet. It collects data about sexual behaviours every 10 years. In 1990, 2000, 2010, in the latest survey the proportion of LGB respondents has increased 3-5 fold. My assumption is not that there are more LGB people in the population than there were 20 years ago but rather people are more comfortable in disclosing. They also use computer aided design to ask these questions so the interviewer does not see the responses.

Most surveys that do investigate LGBT identity are more likely to be conducted by universities or by research consultancies. These surveys are successful in recruiting LGB people, although they need to do more to address the exclusion of trans people.

Julie

On a more general level, people are not likely to fully report their sexual behavior, whether they be members of sexual minorities or not.

Generally speaking, survey research that involves questions about "sensitive issues" like sexuality, but also criminal histories or stigmatized political views, tend to be affected by social desirability bias. Fearing stigmatization or judgement, respondents tend to pick answers that they hope the researcher will approve. In more serious circumstances, when the activity the survey asks about is illegal, for example, respondents may not want to reveal behaviors if they fear that such information could come back to hurt them. This recent article discusses an interesting case related to social desirability bias in survey questions about sexuality in Malawi: http://www.ncbi.nlm.nih.gov/pubmed/23584464

To attempt to get more accurate reporting of sexual behavior, some new experimental models have been tried. For instance, the "unmatched count technique" allows respondents to answer sensitive questions without explicitly saying they engaged in a particular stigmatized behavior. Instead, the questions ask respondents to say how many, out of a list of behaviors, they have engaged in and surveyors can deduce the answers from that. This recent article describes how using this technique yields different results for questions about sexuality than standard questionnaire methods: http://www.tandfonline.com/doi/abs/10.1080/21642850.2014.886957#.Uw-zkv2YTAY

Good question, Matt. Christian's suggestion's about how and why people might refuse a survey altogether, or be reluctant to 'out' themselves in certain circumstances, hold true in Australia too. To get reliable answers from respondents in a survey you need to be 'tellable'; people need to have grounds for trust that you will understand their responses (e.g., not assume they are gay if they admit to any same-sex experience) and that you will use the responses for good purposes.

It is not only a reporting problem, but a participation problem. Research has shown that people who respond to a survey explicitly concerned with sex tend to be more sexually liberal and have wider sexual experience. If people fear questions about sexual orientation, or are conflicted about it, or their identity (what they think of themselves as) does not correspond with their preferences or experience, they are more likely to refuse to take part. In some unsafe settings G/L people will be reluctant to take part. If forced or pressured, they may omit items they do not wish to answer. This is why I think researchers should report missing data rates (don't know, can't remember, refused), as they are an indicator of discomfort with or covert refusal of the question.

In practice, in our national telephone surveys, we have found that older people are more secure in their answers and less likely to be embarrassed by sexuality questions. Young people who are not yet sure of their identity, or not sure what the questions imply, or lack sexual experience, are less comfortable. Bisexual or gender-questioning people are at a disadvantage - e.g. the latter get missed altogether if surveys recruit men and women separately or require unambiguous gender responses up front.

One source of this assumption (that G/L ppl are underrepresented in surveys) is the dismay many GLB activists felt from the 1990s onwards when scientific surveys produced much lower estimates of the proportion of G/L people in the population than Kinsey's from the 1940s, which (apart from being a bit skewed in recruitment, which he acknowledged), over-counted 'homosexuality' by including behaviour that would not nowadays be regarded as gay. His composite measure of homosexuality runs experience and attraction and what we now call sexual identity together. It was disappointing for activists to find that (exclusive or preferentially) G/L people were closer to 1-2% of the population, not 10%+. And many people who have sex with both men and women do not identify as bisexual.

It would be interesting to do a gay community volunteer sample survey and ask the respondents what they said (or would say) in telephone surveys or the Census. We tend to assume that if people take part at all, they say the same in both.

No but it depends on culture and religion they are living,some societies accepts the sexual orientation of guys and lesbian but in conservative societies they hide their orientation,i case of straight most of them hide themselves by cross dressing

We have decades of experience on what individuals indicate their sexual orientation is in the Multicenter AIDS Cohort Study, and in addition to the many explanations and observations above, we have found that the gender match of interviewer and respondent and the race/ethnicity of the respondent are two very important variables. When a female interviewer is used, some men want to seem more masculine, a form of social desirability I guess, and so will indicate that they are heterosexual or bisexual and list many more female partners than same-sex partners. Also, in working with young Black men still living in their home community where homosexuality is highly stigmatized and parents are exhorted by religious leaders and others to throw out their sons if they are gay, very few of the men signing up for a study of men who have sex with men initially report that they are homosexual or even bisexual, more often checking the heterosexual or "other" option. Over time, as they become more comfortable with being in the study and the precautions take to ensure the confidentiality of all their data and to not even send them a message or letter that would indicate that they are participating in a study for MSM, many become more open to discussing their sexuality. However, the one time when revealing sexuality is really important and taken seriously is when someone learns that they are HIV through the study or routine testing outside the study, and are confronted with the fact that there are many services available to Black MSM through the gay community.

@Dr. Ostrow

That is fascinating.

It's certainly tempting to assume that the combination of various enculturated behavioral influences such as shame, fear, gender insecurity, and others would all act as powerful deterrents to disclosing sexual orientation even within a very impersonal survey interaction - but it is important to research these sorts of issues in order to identify patterns we would probably not have even considered without examining data from qualitative inquiry. It's intriguing stuff, but I feel like you should be going a step further with this..

I think it's worthwhile to include an ethnographic interview component to your methodology as well, though. Try assembling groups of research participant samples by approaching gay men in different environments.

Make a few genuine platonic friends in a gay bar and build enough rapport that they will give you an insider's perspective of the issue for one group, then start up 5-15 minute conversations with strangers smoking outside the same bars and ask the same questions... No matter what the data shows, it will add much-needed depth and allow you to contrast the kinds of replies you got from ethnographic subjects who trust you with those you received from virtual strangers during a survey.

A good qualitative, humanist social science research project should try and approach the topic from both a depth- and volume-oriented data collection methodology, and do its research within various environments in order to understand whether the answers you receive are liable to change depending on how comfortable the participant feels during the process. You'd be surprised sometimes!

Phil- You are correct that an ethnographic, as well as qualitative interviews and even focus groups are important in designing survey instruments that will elicit the most useful and valid questions regarding self-revealed sexual orientation, especially at the beginning planning stages of any cross-sectional or longitudinal study. In the case of the MACS, where we had the opportunity to do this sort of pilot work at the beginning of the study in 1983, we are constrained since then that by changing our data collection methods mid-stream, we will cause disruptions in the longitudinal continuum of this observational study. Partly because of that reason, we more and more depend on the actual reported gender of sexual partners than their answer to a single dimensional set of questions that neither captures the full multi-dimensional constructs that together better define sexual orientation nor is appropriate for following changes in sexual orientation and typologies over the course of participation in this very long-term study (30+ years and still going on).

Instead, I suggest that you look at the reports coming out of the family of Health Behavior Research Studies being done every 2-3 years among MSM in 25 States or metropolitan regions collaboratively between the CDC and the Health Departments and Investigators in each State or city. Depending on the emphasis of the MSM cycle of each 3-year grant, it is usually necessary to do the sort of ethnographic and qualitative research to both identify how and where to sample for the desired study population (e.g. younger or minority MSM, drug using MSM, etc.) and the terminology and street language that such men use to identify themselves and members of their socio-sexual networks. I haven't done any systematic review of these reports, but I am pretty certain that one would find differences between metropolitan areas as well as changes across time and specific target subpopulations within the same geographic region regarding the ethnographic/qualitative research findings on the most valid ways to illicit useful information concerning self perceived sexual orientation and behaviors,

This conversation thread is fascinating!

I work in an environment where I'm occasionally called on to help novice researchers with their first surveys. I'm always tempted to suggest trying a likert scale for gender identity and sexual preference responses. Would it provide more representative results than the usual dichotomous choices? The frustrating reality of needing to find a balance between capturing representative data and minimizing the overall length of survey comes in to play & I don't want to complicate the learning process for my novices.. so I use it as an example of how question formation can be easily assumed as obvious by some, and confounding by others.

Since I'm not in a position to test this myself I would be very interested in a study on the topic!

I also think that it really depends on why the question is being asked on the survey in the first place. If the responder can perceive that the orientation and gender questions really do relate to the other parts of the survey, and that by answering them there is some benefit, then I believe they may be more likely to answer. For example, suppose the survey is something like "Improving mentoring relationships with GLBTQ students" - in this case the gender and orientation questions are highly relevant. If it is "The impact of sexual orientation on learning outcomes" the motives for the survey become less clear, and could impact the responses to the gender and orientation questions. I am sure there are better examples, but hopefully you all get the idea

It looks like I never said thanks for all these helpful responses, and I apologize. What a bad participant I've been! We had started a paper on this topic when I posted, and I've spent the past year on it. It will appear in the Am. J. of Public Health sometime this year. I'll post info once it's out. Thanks again. These are great, deep, and thoughtful comments all around.