Request For Withdrawal of Misleading UK Office for National Statistics Data

ONG logo

Monday 7 October 2013

Glen Watson
Director General
Office for National Statistics
United Kingdom


(i)   +44-(0)1633 456179
(ii)  +44-(0)1633 652747
(iii) +44-(0)20 7014 2453

(i) Media Relations, ONS
(ii) Mervyn Stevens
Complaints Officer, UK Statistics Authority

Dear Mr Watson


I write to request you withdraw data published by the ONS last week (3 October 2013) as part of the Integrated Household Survey:

Out Now is a specialist global consultancy that has worked in the area of lesbian, gay, bisexual and transgender (LGBT) research since 1992 and we are very concerned that poor methodology employed in this ONS data has caused the results to be most inaccurate.

In the summary of Key Findings you claim that your research shows that "1.5 per cent of adults in the UK identified themselves as Gay, Lesbian or Bisexual."

This claim is not only inaccurate as a true measure of the prevalence of homosexuality and bisexuality in the United Kingdom, it is dangerous as it is prone to misreporting and may lead to funding problems for organisations focused on the health of LGB people.

The research methodology you have employed is not suitable for the purpose of obtaining an accurate measure of homosexuality and bisexuality in the UK.

That this is so can be seen by reference to research from the United States which uses superior methodologies to arrive at vastly different results in measuring the prevalence of bisexuality and/or homosexuality in the USA.

So far as Out Now is concerned, the Yankelovich MONITOR Study of 1992 reported on in 1993 remains the benchmark standard for methodology when it comes to allowing respondents to identify as gay or lesbian to the researcher. That study was a household survey of US households but their methodology for data collection of such a sensitive metric as respondents' sexual identity was far superior to yours. In this random-sample study, just under six per cent of the US population was found to identify as "gay/lesbian/homosexual".

In the US Yankelovich research, interviews were conducted in respondents' living rooms and the question related to sexual orientation was asked only on the third visit - once a rapport of confidence and trust could have been established between the interviewer and each individual respondent. The Yankelovich research used flip cards which were randomised such that the respondent knew that for the question set including the question on sexual orientation, each interviewer had no way of knowing what specific response an individual respondent had given for this question. In so doing, both researcher and respondent had vastly increased confidence levels that the responses given truly were confidential and not able to be linked back personally to a particular respondent.

In your most recent "Experimental Research" that you reported on last week, your interview subjects were contacted at home, by a stranger they had not previously met and were asked to reveal their sexuality to that person. Your 'randomising' of responses for that question and inviting the respondent to say "stop" when their own sexual orientation was mentioned does nothing to preserve the confidentiality between the researcher and the respondent in relation to what is for many people in the UK a highly sensitive piece of personal information.

If you doubt that your research is flawed by virtue of its poor methodology I suggest you consider the results given for "Don't know / refusal" on the sexual orientation question.

For almost all age groups this result is approaching 4% of the total sample.

Does it not strike anyone at ONS as perhaps a bit 'odd' that this many people might not know their own sexual orientation? If we are therefore to conclude that these respondents are in fact refusing to divulge their sexuality - a not unreasonable assumption we contend, given the clumsy way in which respondents are expected in your research to 'out' themselves to a total stranger - then the only valid conclusion that your research ought to be able to make is:
"Up to 4% of UK people prefer not to divulge their sexuality to a stranger when asked to for an ONS research study".

Of course one could attempt to assert that respondents would be reassured by the researcher that this information is confidential - but that fails to understand a fundamental reality of being lesbian, gay, bisexual or transgender (LGBT) in the UK in 2013: large numbers of people are not 'out' about their sexual orientation - even with many people whom they know very well.

Out Now undertakes a global LGBT research initiative called LGBT2020 and since 2010 this project has sampled more than 15,000 LGBT people in the UK. In the next few weeks that study will be sampling its 100,000th respondent globally on LGBT issues.

What the LGBT2020 research shows very clearly is that for many people, revealing their sexuality is a 'risky business' and they employ a range of tactics in a range of scenarios.

For example, at work in the UK in 2012, the LGBT2020 study shows that only 52% of LGBT respondents were out to all their work colleagues as an L,G,B or T person.

You can read more about this in the "International LGBT2020 Homophobia Report".

That report shows that, of the more than 2,500 UK respondents sampled in the Out Now 2012 study, only 56% of respondents feel able to be out to all their family members.

The reasons for such reluctance are not hard to discern when we consider that 53% of respondents report witnessing issues related to homophobia in their workplaces.

49% of respondents also reported personally experiencing at least one incident of verbal harassment last year due to their sexual orientation or gender identity.

6% were harassed by their neighbours and 15% of respondents experienced personal harassment in their own workplaces because they are LGBT.

Given this background, it perhaps becomes more understandable why your sample reports such a high level of "Don't know/refusal" to answer the question on sexual orientation, when asked by a stranger - a person they have not previously met who knows their phone number and/or their home address.

If you still doubt that your methodology on sexual orientation is to blame for such low reported levels of bisexuality and homosexuality, then you may like to also consider the findings of one of the largest recent peer-reviewed academic research studies on human sexuality:
"National Survey of Sexual Health and Behavior (NSSHB)" - undertaken by the University of Indiana and reported on in 2010. Details are here and you will note that they report:

"While about 7% of adult women and 8% of men identify as gay, lesbian or bisexual, the proportion of individuals in the U.S. who have had same-gender sexual interactions at some point in their lives is higher."

Now we either need to conclude that people in the USA are around five times more likely to be gay, lesbian or bisexual or else we are left with the conclusion that Out Now has reached which is that your research is misleading, inaccurate and dangerous.

In relation to this last point, consider the instances above from the LGBT2020 research by Out Now showing how many LGBT people experienced harassment and discrimination in the UK during the sample period of your own research.

I present - as one example of how inaccurate data in this area can be used against LGB people - this report from the Daily Mail of October 3, 2013:

'1.5% of Britons say they are gay or bisexual: Number hasn't increased in past year despite Cameron's drive for equality
Office for National Statistics found homosexual population has not risen
Number is a quarter of the six per cent claimed by lobbyists and Whitehall'

The data you are responsible for publishing is not fit for purpose and risks being used by those who may not 'like' gay, lesbian or bisexual people to undermine their rights. It may be used similarly to seek to reduce funding for government and NGO initiatives focused on LGB people in the UK.

For this reason, on behalf of Out Now, I request the immediate withdrawal of this data - collected in both 2010 and 2012 - from your reported statistics.

I look forward to your urgent response.

Yours sincerely,

CEO, Out Now