Answering questions about animal behaviour and evolution requires choosing the right analytical tool for the job. Across my research I have drawn on a wide range of statistical and computational methods — each selected because it was the most appropriate way to extract insight from a particular kind of data. This page gives an overview of those approaches, grouped by what they are designed to do.
Social Network Analysis
Animals rarely interact at random — they form structured webs of relationships. Social network analysis maps who interacts with whom, and how often, revealing patterns invisible in simple averages. I have used it to study crickets, dolphins, spiders, and squirrels, among others.
Exponential Random Graph Models (ERGMs)
A more rigorous cousin of basic network statistics. ERGMs test whether the patterns we see in a social network — such as cliques or hierarchy — are more pronounced than chance would produce, while accounting for the fact that connections in networks are not independent of each other.
Stochastic Actor-Oriented Models (SAOMs)
Social networks change over time — friendships form and dissolve. SAOMs model the dynamics of networks, asking what drives an animal to change its social ties. They treat each individual as an agent making decisions, capturing the evolving nature of social structure.
Multilayer Network Analysis
Animals interact across multiple contexts simultaneously — they may fight, mate, and forage with different partners. Multilayer networks stack these different types of interaction into a single framework, revealing how behaviour in one context shapes relationships in another.
Mixed Effects Models
Most biological data has structure — repeated measures on the same individual, individuals nested within groups, groups studied across years. Mixed effects models account for this layered structure, separating genuine biological signals from artefacts of study design.
Animal Models
A specialised mixed model that uses pedigree or genomic information to decompose variation in a trait into genetic and environmental components. This is the standard approach for estimating heritability — how much of the variation in a trait is passed from parent to offspring.
Indirect Genetic Effects (IGE) Models
Your genes affect not just your own traits, but also — through your behaviour — the traits of those around you. IGE models quantify this social transmission of genetic influence, and I have shown they can substantially change predictions about how fast traits evolve.
Multilevel Selection Analysis
Natural selection can act on individuals, but also on the groups they belong to. Multilevel selection analyses partition the total force of selection into within-group and between-group components, revealing when group membership matters for evolutionary change.
Meta-Analysis
A single study can only say so much. Meta-analysis pools results across many independent studies to estimate the overall strength and consistency of an effect. I have used this to ask broad questions — such as how large indirect genetic effects are across the animal kingdom.
Multi-Analyst Studies
Different researchers analysing the same dataset can reach different conclusions. I have contributed to large collaborative projects examining how much analytical variation exists across researchers, and what this means for the reliability of published findings.
Repeatability Analysis
For a behavioural tendency to be meaningful — and evolvable — individuals must behave consistently over time and across contexts. Repeatability analysis quantifies this consistency, providing a ceiling estimate for heritability and a test of whether “personality” is real.
AI-Assisted Tracking
Modern computer vision tools can track individually marked animals automatically through video footage, generating fine-grained data on movement and interaction at a scale impossible by hand. I have used and validated these approaches for studying cockroach behaviour, in particular identifying when social interactions occur.
Survival & Senescence Modelling
How long animals live, and how their performance changes with age, requires specialised statistical tools. I have used survival analysis and age-structured models to study ageing in wild insect populations observed over many years.
Agent-Based & Complex Systems Modelling
Some questions are too complex to answer with data alone. Agent-based models simulate populations of individuals following simple rules, and examine what collective patterns emerge. I have used these to explore how individual behaviour scales up to group-level phenomena.
Theoretical Quantitative Genetics
Mathematical models of inheritance allow predictions about evolution that go beyond what any dataset could directly test. I have developed theoretical frameworks showing how social interactions change the expected rate and direction of evolutionary change.
DNA Methylation Analysis
Epigenetic marks — chemical modifications to DNA that do not change the underlying sequence — can influence gene expression and may even be heritable. I have profiled genome-wide methylation patterns in social spiders to explore how infection alters these marks. Additionally, Hamish William’s PhD project is exploring how beadlet anemones might use DNA methylation to respond to oil pollution