Analysing animal social network dynamics: the potential of stochastic actor‐oriented models

Abstract

Summary1. Animals are embedded in dynamically changing networks of relationships with conspecifics. These dynamic networks are fundamental aspects of their environment, creating selection on behaviours and other traits. However, most social network-based approaches in ecology are constrained to considering networks as static, despite several calls for such analyses to become more dynamic.2. There are a number of statistical analyses developed in the social sciences that are increasingly being applied to animal networks, of which stochastic actor-oriented models (SAOMs) are a principal example. SAOMs are a class of individual-based models designed to model transitions in networks between discrete time points, as influenced by network structure and covariates. It is not clear, however, how useful such techniques are to ecologists, and whether they are suited to animal social networks. 3. We review the recent applications of SAOMs to animal networks, outlining findings and assessing the strengths and weaknesses of SAOMs when applied to animal rather than human networks. We go on to highlight the types of ecological and evolutionary processes that SAOMs can be used to study. 4. SAOMs can include effects and covariates for individuals, dyads and populations, which can be constant or variable. This allows for the examination of a wide range of questions of interest to ecologists. However, high-resolution data are required, meaning SAOMs will not be useable in all study systems. It remains unclear how robust SAOMs are to missing data and uncertainty around social relationships. 5. Ultimately, we encourage the careful application of SAOMs in appropriate systems, with dynamic network analyses likely to prove highly informative. Researchers can then extend the basic method to tackle a range of existing questions in ecology and explore novel lines of questioning.

The application of statistical network models in disease research

“…The data requirements, as well as details on tests for model convergence, goodness‐of‐fit and significance, are provided elsewhere (Ripley, Snijders & Preciado ; Ilany, Booms & Holekamp ; Fisher et al . ). However, we highlight two important considerations of direct relevance to disease research.…”

“…Stochastic actor-oriented models (SAOMs) use an individualbased approach to model how network structure changes through time, and can link these changes to structural features of the network, individual traits or dyadic covariates (Snijders, Van de Bunt & Steglich 2010;Fisher et al 2017). Model terms (structural terms, and individual or dyadic covariates) can be used to explain both the rate that an individual has an opportunity to change to its network position (the 'rate' function) and the probability that it does so when the opportunity arises (the 'objective' function) (Snijders, Van de Bunt & Steglich 2010;Ripley, Snijders & Preciado 2011).…”

“…However, similar to ERGMs, it is possible to control for structural features in interaction-or contact-based data using covariates, e.g. distance effects or shared group effects (Fisher et al 2017). SAOMs can currently model only binary or ordered networks, so are best used in cases where the presence/absence of an edge is more informative than its weight, or when network snapshots are constructed over relatively short time windows (Fisher et al 2017).…”

“…distance effects or shared group effects (Fisher et al 2017). SAOMs can currently model only binary or ordered networks, so are best used in cases where the presence/absence of an edge is more informative than its weight, or when network snapshots are constructed over relatively short time windows (Fisher et al 2017). However, being able to incorporate ordered networks does at least enable relationships of different strengths to be modelled separately (see http://www.stats.ox.ac.uk/ $ snijders/siena/RscriptSiena Ordered.R), which may be important for particular diseases or social systems.…”

“…SAOMs are implemented in R using the package RSiena (Ripley, Boitmanis & Snijders 2013). Models are best constructed in a stepwise manner (see Supporting Information), starting with basic structural terms and adding in more complex structural terms, and then behavioural terms, once the current model converges and fits the data at each step (Ilany, Booms & Holekamp 2015;Fisher et al 2017). The data requirements, as well as details on tests for model convergence, goodness-of-fit and significance, are provided elsewhere (Ripley, Snijders & Preciado 2011;Ilany, Booms & Holekamp 2015;Fisher et al 2017).…”