Download this episode
In Episode 9 of Series 3 of the DIAL Podcast, Professor Andrew King and Matt Hall from DIAL’s CILIA-LGBTQI+ research programme discuss their work exploring how Agent Based Modelling (ABM) can contribute to the study of LGBTQ lives, and conversely, how theory and insights from LGBTQ studies can inform the practice of ABM.
Further links
- Queer(y)ing Agent-Based Modelling: An example from LGBTQ workplace studies is a DIAL Working Paper
Transcript
In Episode 9 of Series 3 of the DIAL Podcast, Professor Andrew King and Matt Hall from DIAL’s CILIA-LGBTQI+ research programme discuss their work exploring how Agent Based Modelling (ABM) can contribute to the study of LGBTQ lives, and conversely, how theory and insights from LGBTQ studies can inform the practice of ABM.
Queer(y)ing Agent-Based Modelling: An example from LGBTQ workplace studies is a DIAL Working Paper
Christine Garrington 0:00
Welcome to DIAL a podcast where we tune into evidence on inequality over the life course. In series three of the podcast, we’re discuss emerging findings from DIAL research. For this episode, I’m joined by Professor Andrew King and research fellow Matt Hall from the University of Surrey, to discuss new research, exploring how Agent Based Modelling (ABM) can contribute to the study of LGBTQ lives, and conversely how theory and insights from LGBTQ studies can inform the practice of ABM. Before Matt walked us through what they did, I asked them both to explain the backdrop to the research.
Andrew King 0:35
When we were putting the project together in the first place, I’d become aware of the potential of Agent Based Modelling to explore ideas and theories and policies, especially on topics with populations where there were quite different data sets available which could be combined and synthesised. And in the case of LGBTQ people. There are quite a lot of smaller qualitative studies using interviews and focus groups and other qualitative data. And then there are some surveys with quantitative data. There are also areas where data is very limited and missing, particularly in relation to bisexual, trans and queer people. Hence, Agent Based Modelling really offered the CILIA project a new way of addressing these issues and trying something quite different, and quite novel, but there is much more to it than that, as Matt will go on to explain.
Matt Hall 1:43
Yeah we’re also keen to address what’s previously been termed methodological binarism within the field. So as Andy’s just mentioned, there are numerous smaller qualitative studies and some larger quantitative data sets available, but there’s very little conversation going on between the two of them. So each of these types of study tends to form its own stream, and then with its own limitations, and we identify this as a particular barrier for forwarding intersectional insights within policy domains and responding to how unique inequalities can emerge at the intersections of different social identities. So, where the quantitative research isn’t engaging enough with intersectionality, or smaller sub populations within LGBTQ qualitative research is then limited in its ability to demonstrate the wider implications or significance of the details that it focuses on. So, it needs to be able to demonstrate the cumulative impact of those details like intersectional experiences. And although it’s complete misreading of intersectionality theory, there’s this tendency for conversations about intersexual inequality, particularly in public and policy spheres, is to treat inequalities as simply additive to where experiences of inequality of just the sum of those afforded by each of the persons identities.
Christine Garrington 3:10
Yeah, that’s really interesting now you propose something that you describe in the paper as a double querying approach which clearly has a bit of a double meaning but I wonder if you can just explain to us what exactly you mean there?
Matt Hall 3:23
Well, although this paper was mainly written with LGBTQ scholars in mind as our audience were really keen to not just make this a paper encouraging the uptake of Agent Based Modelling within the field. We wanted just as much to explore what insights the rich theoretical traditions of LGBTQ scholarship can bring to ABM as a developing methodology. So many of the normative methodological practices and judgments surrounding Agent Based Modelling are shaped by the academic fields that most frequently use it like artificial intelligence research, ecology and epidemiology, and we wanted to clearly present ABM as more of a tool that’s flexible for the needs of different disciplines. So kind of demystifying it for many of its more quantitative oriented applications. And in doing this, we were particularly interested in exploring its compatibility with intersectional and queer perspectives, which are currently quite dominant paradigms within LGBTQ research. So although the paper introduces our own LGBTQ workplace model, and briefly summarises some of the results we found interesting. It is intended much more as a methodological paper, demonstrating what Agent Based Modelling can bring to LGBTQ research, and in turn, some of the challenges we had to negotiate in integrating it with intersectional and queer insight.
Andrew King 4:47
Yes, we were very much wanting to speak to too often quite distinct audiences with this paper, and try to create space for thinking and dialogue, and we hope that this podcast will help with that sort of knowledge sharing and provoke new dialogue and engagement.
Christine Garrington 5:09
So Matt, you’re going to walk us through the detail of the research and then we’ll get some final reflections from Andrew. So can we talk a bit more about what all this looks like in day-to-day life you use the example of career progression in the workplace. What was it here that you actually examined?
Matt Hall 5:23
So, as well as intersectionality, the wider CILIA project is particularly focused on life course inequality. And to us, this means exploring how inequality emerges at different key stages in the life course and how these then intersect. So for example, someone may enter the workforce with a good education and strong mobility, and then face x amount of discrimination within their workplace. So we could then simply look at their final career outcome as a product of their ability and exposure to discrimination over time. But people’s careers, just like the rest of their lives don’t really fold out like that in such a linear way. There are key stages where being exposed to discrimination may have a larger impact than other stages, like when seeking a promotion or when aligned with other non-work-related experiences in someone’s life, such as undergoing a gender transition, experiences of racial discrimination or in the case of bisexual employees, even the gender dynamics of their current relationships. Similarly, employees have a bounded agency in how they respond to and protect themselves from experiences of discrimination at work. Those able to move to less discriminatory workplaces. Some may be able to create straight or cisgender aliases for themselves, while others. Others try to avoid situations where gender or sexuality are discussed at all, and others still may even feel empowered to make themselves visible, and even instigate change within their workplaces. And then each of these strategies that someone chooses has its own consequences for that individual, but also cumulatively on their workplace environment. Of these options presented to LGBTQ people are also often contingent on the alignment of many of those other workplace and life course experiences that we’ve just spoken about. So rather than just a simple model predicting someone’s final career outcome based on their assets and demographics. We wanted to explicitly model career progression as this process. Specifically, complex linear process involving interactions between people and their workplaces, various degrees of autonomy and outcomes that are somewhat path dependent. We also wanted to explore the extent to adding complexity, particularly intersectional would impact overall outcomes, in essence, demonstrating how small qualitative details, often omitted from simpler linear models can have significant affect in terms of quantifiable outcomes.
Christine Garrington 8:04
So imagine this was no straightforward process how exactly did you go about it?
Matt Hall 8:10
So because of conceptualising career progress as a complex and interactive process using simulation methodology, in this case Agent Based Modelling just made complete, complete sense to us. However, we did still need to significantly simplify, how we represented career progression in the model. So in order to keep the model simple enough to explore and to understand the causal mechanisms behind any interesting emergent behaviours, we ended up representing three idealised stages in an individual’s career – so school to work transition, a mid-career transition, and then a transition into retirement. And each LGBTQ citizen in the model, so our agents, were all distributed an LGBTQ status, so whether they were lesbian, gay, bisexual, male, trans, female, etc, and an ethnicity and social class, and then also with other characteristics like ability and access to social capital being proportionately distributed according to these identities and based on existing survey data. Then during each time step in the simulation, or every year, each agent has a probability of entering one of these career transitions based on their age. And the success of each of these transitions are based on a number of factors such as their ability for success of their last transition experiences of discrimination, and in some scenarios, we also included social capital, and individual’s identity management practices as additional factors, and we were then able to explore the cumulative impact of these career transitions on overall career outcomes for individual agents, across the whole population of agents, for particular identity groups, and across the different intersections of those identity groups as well.
Christine Garrington 10:00
And what did you see from the different scenarios that you modelled?
Matt Hall 10:04
Firstly, when exploring the impact of the different model scenarios on workplaces, and their discriminatory practices, we found that increasing complexity and the agency of LGBTQ workers to for example move workplaces if unhappy, and manage their identity in different ways, didn’t have that much impact on reducing the discriminatory practices of the average workplace. This tended to be fairly low and stable over all the model runs, it did significantly impact the discriminatory practices of most discriminatory workplaces. Although interestingly, this didn’t necessarily correspond with improved career equality for the agents themselves, in fact it tended to be the simplest scenarios that saw the largest improvements to LGBTQ career equality over time. With career equality here being the correlation between each individual’s ability, and their career outcome.
Christine Garrington 10:58
Okay and there were differences within the group that you were looking at also. What did you see that seemed important there?
Matt Hall 11:03
Yeah, so, generally, we saw that the more complex, or the more detailed the theory that we implemented the more career inequality we observed between different LGBTQ strands. So, in the simplest scenarios where all agents experienced different levels of discrimination during their career transitions, based on LGBTQ ethnic and class status. In these, the different identity groups would end up with different career outcomes, but based purely on these initial levels of discrimination. The actual trajectories would be more or less the same between them. So career equality steadily increasing over time for everyone, and at a similar rate. However, over the different scenarios, as we incrementally increased complexity in the career progression process like adding in social capital as a factor mediating people’s capacity to move to better workplaces, and their identity management practices and intersectional variations, we began to see different trajectories emerging through equality between each of the different LGBTQ strands and between the intersections, within each of those strands.
Christine Garrington 12:13
And other factors such as someone’s background or ethnicity, they were also important weren’t they?
Matt Hall 12:18
Yeah, so for example, in the most complex scenarios, whilst career equalities steadily increased over time for the overall agent population, when we then aggregate down – desegregate down to the non-white working class trans and queer agents we observed the complete opposite trajectory happening. So a fairly dramatic reduction in this case in career equality, over time, and this highlighted to us firstly how just fairly small intersectional differences in the career progression process cumulative – can accumulate significantly over the life course in somewhat less predictable ways than linear models would usually predict. So, a minor reduction in an individual’s behavioural opportunities like moving to a better workplace or feeling confident enough to disclose their sexual or gender identity in the workplace can set the same individual on completely different trajectories in terms of career outcome over time. And secondly, it highlights the importance of aggregating down to much smaller intersections of population in our models, something that traditional quantitative methods struggle with due to losing statistical significance when aggregating down to smaller samples. And it’s been an ongoing problem for research on LGBTQ populations in general, let alone when we start introducing intersections with other identities like ethnicity, and social class.
Christine Garrington 13:46
Such a really important and major piece of work from you both. I wonder what you take away from this and how it can inform policymakers and employers who are interested in reducing inequalities for LGBTQ workers.
Matt Hall 14:00
in terms of informing policymakers, our model hopefully demonstrates the importance of considering the impact of intersectional experiences on career inequality. So currently, both wider equality legislation across Europe and more localised workplace policies, tend to be single issue in their approach to social inequalities. For example, they’ll have one policy for addressing LGBTQ issues, one for gender, and another for race, disability, etc. And because of this they very rarely address some of the more complex experiences occurring at the intersection of these identities, there seems to be this underlying assumption, that simply addressing the concerns of each of an identity category separately will automatically meliorate all of the problems emerging at the intersections of them. And as we mentioned earlier, this comes from assuming that inequality is additive or a simple sum of each experience of inequality, whereas what we’ve hopefully demonstrated with our model is that inequality synergized in much more complex ways than this, and that we need to encourage more intersectionally aware policymaking. For example, this may be race policies that acknowledge that many ethnic minority individuals are also gay, lesbian, bisexual, trans, and likewise LGBTQ policies that explicitly address how some LGBTQ employees will also be from ethnic minority backgrounds, different social class backgrounds, and have different abilities and neuro diversities for example, and that a one size fits all approach to LGBTQ equality, may not even begin to start addressing those individual needs at all.
Andrew King 15:38
Yeah, we don’t really see what we’ve done is purely a thought experiment or all about advancing theory and methodology although those things are important in the way we’ve described. We think that what we’re doing can help to shape policy. And what we’re aiming to do a little later in the project is to really engage, policymakers, and employers in these conversations.
Christine Garrington 16:07
Queer(y)ing Agent-Based Modelling: An example from LGBTQ workplace studies is a DIAL Working Paper by Matt Hall and colleagues. It’s part of DIAL’s CILIA-LGBTQI+ project, which is providing cross cultural evidence for the first time ever on life course inequalities experienced by LGBTQI+ people. Thanks for listening to this episode of our podcast which is presented and produced by Chris Garrington and edited by Elina Kilpi-Jakonen.