I’ve just attended the 9th National LGBTI Health Conference in Canberra. The conference organisers had a very progressive approach to communicating with delegates – for the few months leading up to the conference they sent out short announcements (blog posts) about the papers to be presented, along with the more usual updates with delegate information. We also received a post called, “A safe and inclusive conference”. Not something I’ve ever received from a conference but very much appreciated. It has lots of useful thought provoking and anxiety relieving advice. From questions about how we re-tell the personal stories we hear at the conference, through inclusive language to tips on how to avoid misgendering. Honestly, this is useful life advice.
One point they raised resonated with me. In a section on ‘LGBTI’ and inclusion they said:
“Deliberateness: How can we make sure that we move from habitually using all five letters to earning each of them? Is it appropriate to use all five letters or does the topic we are discussing apply more specifically only to some of these populations and need rethinking for some populations?”
Recently, for a project on LGBTI smoking, my colleague Israel Berger and I reviewed published evaluations of smoking cessation interventions for LGBTI individuals (19 studies). We found that:
* All studies included gay men
* About two thirds of studies used general terms like ‘LGBT’ but didn’t necessarily include every group.
* About two thirds of studies mentioned bisexual participants as targets/participants, but there was insufficient reporting of bisexual status. Indeed, several studies used the term LGBT or LGB but then only referred to lesbians and gay men effectively erasing bisexual people.
* About two thirds of studies were nominally open to women, but only a quarter of those studies had women participants (of those that reported gender at all)
* A quarter of studies mentioned trans people, but trans people only represented 3% of participants (of those that reported trans status).
* None of the reviewed studies targeted or reported intersex participants.
So despite two thirds using general terms suggesting their intervention was developed and evaluated for LGBT people, few had earned this terminology. The problem here should be obvious – it looks like we know quite a bit about how to develop and deliver smoking cessation programs to L+G+B+T people. When in fact, we’re on shaky ground for most of these letters.
At the conference I heard several examples of researchers claiming LGB/T/I when their sample was no where near that. I wonder why we do this. And I say “we” deliberately as I know I have done/do this. The conference organisers’ interpretation of this practice is, habit. And so they frame their advice in terms of being deliberate, mindful of the language you use (I appreciate the list of dos in their guidance where others would have a list of do nots). But I wonder if we also over claim inclusiveness because we feel our research should be applicable to all the letters. Even if in practise we can make no such claim. Or do we think the gesture to inclusiveness is sufficient? Or (worse) do we think the whatever we find for some letters will apply to them all?
We had an interesting discussion at the conference about how to earn the letters. For example, should we design our surveys so all the members of our LGBTI communities feel recognised and able to participate in ways that capture their experiences? A good idea but is it enough? The two surveys I’m involved with seek to do this, we ask a question about trans status and a separate question about intersex status (the letters I think that are most commonly claimed but not earned).
What if I don’t do any targeted recruitment? To go back to the review of smoking cessation studies – most were nominally open to women but had low numbers that suggest to me a failure to engage and/or failure to provide culturally safe programs for women. So is saying trans and intersex people are welcome to do my survey – look! I wrote questions – earning the T or the I? I’d find this position hard to stand behind.
What if I have the separate questions but my question responses don’t adequately reflect the diversity of trans or intersex people’s lived experiences? Have I earned it? Hard to argue yes. I’ve had some feedback that one of my surveys does this so my colleagues and I will think carefully about the claims we make about the people our research findings reflect.
What about if during analysis I collapse the beautifully crafted and community-consulted question responses because the cell sizes are too small to be statistically meaningful. Does this make the original attempt tokenism? I am worried that it might be. Yet reporting the % of trans and intersex identifying people but doing no further analysis is what I do in my survey research. I feel uncomfortable but I’m not sure what else we can do.
At the conference wrap-up, rapporteur Terence Humphreys (from Twenty10) said” ‘We are entering a new and nuanced era of deliberately engaging with peoples bodies, genders, sexualities, +identities”. This echoes some work colleagues and I did in relation to same-sex attracted women. We argued that there are important and meaningful differences under the ‘same-sex attracted’ umbrella and this demands a nuanced approach to health promotion. I think this is the better response to the conference organisers’ call for deliberateness. Claim the letters that do reflect the population your research is about, be transparent about the boundaries. And own who is missing.
It’d be great to hear about how you earn the LGBTI in your research…
Great article Julie. Rather than collapsing responses due to small cell sizes and risk losing responses by trams or intersex people, why can’t you use a stratified sampling approach during recruitment to ensure a large enough sample size within each group? This would demand a more targeted approach to recruitment but the benefits in terms of truly representing all five letters are clear.
This is really helpful on lots of counts, especially with regard to our proposed survey. I am interested in the idea of “stratified sampling” suggested by Zoe. Can you please explain what this is and how it might be appleied in our potentially small sample here in Mongolia? Thanks, K
Let me preface this by saying I’m not a statistician 🙂
My understanding of stratified sampling in quantitative research is that you would say X% of the population is trans*, X% is intersex, X% is lesbian, x% is bisexual, etc and then use relevant recruitment strategies for each strata to ensure that the proportions in your sample reflect these population proportions. This is advantageous because you are more likely to reach your minority groups when you are deliberately thinking about how to reach them (rather than just casting the net and hoping they respond which we know does not work).
A potential problem is that (I think) you are supposed to have mutually exclusive strata. I think we can all agree that the letters in LGBTI have a lot of overlap, not least because they are representing sexuality and gender experiences. Maybe you could stratify for sexuality, and trans* or not, and intersex or not??
The other problem is that if the % in the population is already very small (I think <2% of babies are intersex) then stratified sampling won't help. You still end up with small numbers if your sample is not massive, and let's be honest, massive is pretty rare in a community survey. I think you'd need power calculations to work out what is large enough to avoid collapsing cells. And that is the limit of my knowledge!
LikeLiked by 1 person
Which is not to say I don’t think we should totally do targeted recruitment to strata we know are typically under-represented so at the very least people are being captured!
And I am sure there are fancy statistics (modelling?) you can do to deal with the errors produced by small cell sizes…