Bayesian estimation of the network autocorrelation model

详细信息    查看全文

• 作者：Dino Dittrich^a ; ^{dino.dittrich@gmail.com} ; Roger Th.A.J. Leenders^b ; Joris Mulder^a
• 关键词：Network autocorrelation model ; Bayesian inference ; Jeffreys rule prior ; Informative prior distribution ; Frequentist coverage
• 刊名：Social Networks
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：48
• 期：Complete
• 页码：213-236
• 全文大小：1509 K
• 卷排序：48

文摘

Jeffreys prior is derived for the network autocorrelation model. An empirical informative prior for the network effect parameter is provided. Efficient MCMC techniques to sample from the resulting posteriors are developed. Simulation results suggest that the Bayesian estimators outperform the ML one.