The Emergence of Opinion Leaders in a Networked Online Community: A Dyadic Model with Time Dynamics and a Heuristic for Fast Estimation

Abstract

W e study the drivers of the emergence of opinion leaders in a networked community where users establish links to others, indicating their "trust" for the link receiver's opinion. This leads to the formation of a network, with high in-degree individuals being the opinion leaders. We use a dyad-level proportional hazard model with time-varying covariates to model the growth of this network. To estimate our model, we use Weighted Exogenous Sampling with Bayesian Inference, a methodology that we develop for fast estimation of dyadic models on large network data sets. We find that, in the Epinions network, both the widely studied "preferential attachment" effect based on the existing number of inlinks (i.e., a network-based property of a node) and the number and quality of reviews written (i.e., an intrinsic property of a node) are significant drivers of new incoming trust links to a reviewer (i.e., inlinks to a node). Interestingly, we find that time is an important moderator of these effects-intrinsic node characteristics are a stronger short-term driver of additional inlinks, whereas the preferential attachment effect has a smaller impact but it persists for a longer time. Our novel insights have important managerial implications for the design of online review communities.

Forgotten Third Parties: Analyzing the Contingent Association Between Unshared Third Parties, Knowledge Overlap, and Knowledge Transfer Relationships with Outsiders

“…For approximately 20 million observations there were only 20421 instances where the dependent variable equaled 1. Hence, to deal with the computational requirements of our dyad-level model with time-varying covariates, we estimated our model with the recently proposed Weighted Exogenous Sampling with Bayesian Inference (WESBI) (Lu et al 2013).…”

“…By combining these two sets of observations, we constructed a much smaller dataset ("sampled data"). And then, we used the weighted log-conditional-likelihood function for Bayesian inference over our sampled data (Lu et al 2013). The intuition behind the weighted log-conditional-likelihood is to weigh each sampled observation by the population elements it represents in order to make the choice-based sample simulate a random exogenous sample.…”

"O Que Gerencio e de Quem Dependo?": Determinantes da Ação de Blogueiros

“…Dessa forma, o esforço de gerenciamento pelo blogueiro dos recursos dos quais o blog depende tende a ser decrescente à medida que a página adquire atratividade e se torna menos instável, uma consequência do padrão ricos ficam mais ricos (Barabási & Albert, 1999;Lu, Jerath, & Singh, 2013;Newman, 2003;Sundararajan et al, 2013).…”

“…Entretanto, a perspectiva de Barabási e Albert (1999) não discute a possibilidade da superação dessas barreiras pelo gerente do blog, estando intrinsicamente estabelecido que os ricos ficam mais ricos, de modo que o ator altamente atrativo tende a ser atrativo para sempre, sem precisar exercer esforços substanciais na aquisição de recursos, ainda que outros atores centrais surjam e coexistam na rede (Freeman, 2004;Harrigan, Achananuparp, & Lim, 2012;Lu et al, 2013). Nesse contexto, apesar de diversos estudos (e.g., Ackland & O'Neil, 2011;Chau & Xu, 2012;Yang & Counts, 2010) corroborarem que os blogs são estruturados sob o mecanismo de conexão preferencial, é natural se perguntar de que maneira então é possível que blogueiros pobres cumpram o desafio de se tornarem ricos, já que a rede é dinâmica e a todo o momento blogs são conectados, reconectados e desligados, abrindo espaço para que blogs menos atrativos assumam posições de páginas altamente atrativas ao longo do tempo.…”

“…São necessárias, portanto, mais investigações que discutam a validade da perspectiva de conexão preferencial em ambientes instáveis, heterogêneos e em que há dependência de recursos, que consequentemente exigem do blogueiro um gerenciamento de necessidades e contingências para a sobrevivência da página, já que o favoritismo imposto pela conexão preferencial deveria, em tese, limitar o sucesso de blogs de baixa atratividade (Barabási & Albert, 1999;Harrigan et al, 2012;Lu et al, 2013).…”