I’m dedicating this post to my brilliant and amazing (and patient and tolerant) dissertation co-chairs, who are co-authors on this:
Risk, Individual Differences, and Environment: An Agent-Based
Modeling Approach to Sexual Risk-Taking. (Also known as: Emily is a big giant nerd.)
(EDIT: Download the PDF here.)
(This is official as of today!)
I defended my dissertation in August of 2006, having spent the prior 18-24 months preparing it. It took 5 years to get it published.
Part of why it took so long is that I lose at least 3 months out of each year to seasonal mood issues that seriously disrupt my ability to work – like this, for example. And also I left the state, so we had to collaborate from a distance, which is surprisingly difficult. And then there’s the fact that I have a more-than-full-time job, and blah-dee-blah.
And then there’s the fact that no one had any idea what to do with it, because it’s an agent-based model, and no one has ever used ABM in sex research before.
It really is new stuff. People have no idea how to feel about it. It was a struggle to write the paper in a way that made it clear to people who have never even heard of computer modeling why anyone would bother.
I believe in agent-based modeling as an approach to public health research in general and sexual health research in particular, because it has the power to model immensely complex and heterogeneous populations.
What is ABM? It’s like a computer game. It’s a little virtual world full of critters. You give the critters rules to live by and an environment in which to live, and then you watch what happens. You can also add in generational factors that allow for evolution of strategies.
Whenever we have an HIV vaccine (or cure), ABM can provide dissemination models that take into account not only basic infrastructure like roads and electricity, but also sociocultural factors like stigma and vaccine resistance.
To study the evolutionary origins of human sexuality, which is tremendously difficult because our sexual cultures leave no fossil record so no one REALLY KNOWS what the hell our early ancestors did with and about sex, we can try multiple models bared on various hypotheses and essentially reverse engineer the origins of human sexuality.
To study the development of childhood sexuality, something that hugely complex due to human subjects/informed consent issues, we can model behaviors and developmental phases that it might be unethical to observe or induce in real life.
And so much more. I really believe ABM is a powerful tool. In particular I think it’s important because it allows us to witness influence ACROSS LEVELS OF ANALYSIS. It can show us how individual-level rules and behaviors can give rise to various patterns at the population level, and vice versa. Which is what my paper did.
My own model was almost painfully simple. Little heterosexual critters (goils and bois) in a landscape-free environment given sex-specific but heterogeneous sexual motivation profiles (SMP) and told to go mate! Then I dropped an HIV-like disease in the system and watched how the infection spread through the system. The independent variable was the rpopulation-level distribution of different SMPs.
In some trials, there were loads of agents with high-risk profiles, and in others there were loads of low-risk profiles, and in others there was a normal distribution of risk profiles (which is what real life is like).
The hypotheses were : (1) that we could make a model that behaved like real life and (2) the movement of the infection through the population would vary depending on the distribution of SMPs in the population.
Even my committee was surprised.
Anyway, I’m terribly proud of my little goils and bois, and I’m dissolving with gratitude to the best committee anyone ever had.