TY - JOUR
T1 - Homophily dynamics outweigh network topology in an extended Axelrod's Cultural Dissemination Model
AU - Dinkelberg, Alejandro
AU - MacCarron, Pádraig
AU - Maher, Paul J.
AU - Quayle, Michael
N1 - Publisher Copyright:
© 2021 The Author(s)
PY - 2021/9/15
Y1 - 2021/9/15
N2 - Personal social networks reveal potential sources of dyadic social influence. Social influence is picked up as a main principle of Axelrod's model of cultural dissemination. Even though social influence is performed via social networks, the model is generally just run on a regular lattice instead of more complex network topologies. In this paper, we analyse a concurrent extension to Axelrod's model for opinion-based groups, and explore the performance of changing the network topology. Our objective is to seed the Axelrod model with attitudinal survey data as an empirical data application. In the model, the culture is a set of features which in turn is defined by a set of traits. Respectively, in survey data, the attitudes are captured by items with a fixed set of response options. The direct correspondence of the structure of survey data to the model makes it an ideal candidate. Here, we simulate and analyse the extended Axelrod model to explore its dynamics and outcomes, with the standard Axelrod model results serving as a benchmark. As well as the lattice, which the Axelrod model is usually simulated on, we test other network topologies. The conducted simulations explore the parameter space for the uniformly distributed models, and draw parallels between the results, when applying it to an empirical, attitudinal data set. After assessing the level of impact of the network structures, we conclude that there is almost no influence of the underlying network structure on the macro level outcomes. The reason seems to be that the homophily structure among the individuals outweighs the impact of the network topology in the long run simulations. Under the premise, that the number of features is higher than the number of traits and that the system size is limited, the extended Axelrod model can be used to simulate attitudes from a survey — without specifying the underlying network.
AB - Personal social networks reveal potential sources of dyadic social influence. Social influence is picked up as a main principle of Axelrod's model of cultural dissemination. Even though social influence is performed via social networks, the model is generally just run on a regular lattice instead of more complex network topologies. In this paper, we analyse a concurrent extension to Axelrod's model for opinion-based groups, and explore the performance of changing the network topology. Our objective is to seed the Axelrod model with attitudinal survey data as an empirical data application. In the model, the culture is a set of features which in turn is defined by a set of traits. Respectively, in survey data, the attitudes are captured by items with a fixed set of response options. The direct correspondence of the structure of survey data to the model makes it an ideal candidate. Here, we simulate and analyse the extended Axelrod model to explore its dynamics and outcomes, with the standard Axelrod model results serving as a benchmark. As well as the lattice, which the Axelrod model is usually simulated on, we test other network topologies. The conducted simulations explore the parameter space for the uniformly distributed models, and draw parallels between the results, when applying it to an empirical, attitudinal data set. After assessing the level of impact of the network structures, we conclude that there is almost no influence of the underlying network structure on the macro level outcomes. The reason seems to be that the homophily structure among the individuals outweighs the impact of the network topology in the long run simulations. Under the premise, that the number of features is higher than the number of traits and that the system size is limited, the extended Axelrod model can be used to simulate attitudes from a survey — without specifying the underlying network.
KW - Agent-based simulation
KW - Attitudes
KW - Social dynamics
KW - Social influence
UR - http://www.scopus.com/inward/record.url?scp=85107648191&partnerID=8YFLogxK
U2 - 10.1016/j.physa.2021.126086
DO - 10.1016/j.physa.2021.126086
M3 - Article
AN - SCOPUS:85107648191
SN - 0378-4371
VL - 578
SP - -
JO - Physica A: Statistical Mechanics and its Applications
JF - Physica A: Statistical Mechanics and its Applications
M1 - 126086
ER -