Diffusion/Contagion Processes on Social Networks

Abstract

This study models how new ideas, practices, or diseases spread within and between communities, the diffusion of innovations or contagion. Several factors affect diffusion such as the characteristics of the initial adopters, the seeds; the structure of the network over which diffusion occurs; and the shape of the threshold distribution, which is the proportion of prior adopting peers needed for the focal individual to adopt. In this study, seven seeding conditions are modeled: (1) three opinion leadership indicators, (2) two bridging measures, (3) marginally positioned seeds, and (4) randomly selected seeds for comparison. Three network structures are modeled: (1) random, (2) small-world, and (3) scale-free. Four threshold distributions are modeled: (1) normal; (2) uniform; (3) beta 7,14; and (4) beta 1,2; all of which have a mean threshold of 33%, with different variances. The results show that seeding with nodes high on in-degree centrality and/or inverse constraint has faster and more widespread diffusion. Random networks had faster and higher prevalence of diffusion than scale-free ones, but not different from small-world ones. Compared with the normal threshold distribution, the uniform one had faster diffusion and the beta 7,14 distribution had slower diffusion. Most significantly, the threshold distribution standard deviation was associated with rate and prevalence such that higher threshold standard deviations accelerated diffusion and increased prevalence. These results underscore factors that health educators and public health advocates should consider when developing interventions or trying to understand the potential for behavior change.

DamonCentola. How Behavior Spreads: The Science of Complex Contagions. Princeton, NJ: Princeton University Press, 2018. $35.00. Hardcover. pp. 312. ISBN 9780691175317.

“…Conversely, wide bridges, which are multiple redundant connections between people in different neighborhoods or communities, allow for this spread because more reinforcing ties are present. These findings have been echoed in other health behavior research concerning diffusion (Valente & Vega Yon, 2020).…”

Participation in the Cardiovascular Health Awareness Program (CHAP) by older adults residing in social housing in Quebec: Social network analysis

“…The quantitative phase consisted of gathering data to examine each person’s relationships to all others in their own building [ 40 , 42 ]. This methodology involved interviewing at least 25% of all building residents, to ensure centrality measures as in-degree (well suited to identify opinion leaders [ 24 ]), group membership, and total degree (diffusion measure based on opinion leader diffusion model [ 25 , 28 ]), as indicators of network position [ 43 ].…”

“…The qualitative phase was subjected to thematic analysis [ 47 ] performed with NVivo using the deductive/inductive method. Table 1 summarizes the measures used for the analysis [ 24 , 25 , 28 , 40 , 48 , 49 ].…”

“…Identifying the wrong person as a key agent to implement a program can interfere with the program’s benefits [ 12 , 18 ]. Diffusion theory can help explain how new ideas and behaviors spread within communities and among groups [ 27 , 28 ]. In the fields of health education and health behaviors, it has been applied to study how a new idea or practice penetrates [ 27 ] and spreads into a group [ 29 ] through interpersonal communication [ 28 ] and peer leaders [ 30 ].…”

“…Diffusion theory can help explain how new ideas and behaviors spread within communities and among groups [ 27 , 28 ]. In the fields of health education and health behaviors, it has been applied to study how a new idea or practice penetrates [ 27 ] and spreads into a group [ 29 ] through interpersonal communication [ 28 ] and peer leaders [ 30 ]. SNA is thus ideally suited to evaluating the diffusion within a network of behaviors such as attending CHAP sessions [ 13 , 17 ].…”