In Bayesian analysis, the selection of a prior distribution is typically done by considering each parameter in the model. While this can be convenient, in many scenarios it may be desirable to place a prior on a summary measure of the model instead. In this work, we propose a prior on the model fit, as measured by a Bayesian coefficient of determination (R 2 ), which then induces a prior on the individual parameters. We achieve this by placing a beta prior on R 2 and then deriving the induced prior on the global variance parameter for generalized linear mixed models. We derive closed-form expressions in many scenarios and present several approximation strategies when an analytic form is not possible and/or to allow for easier computation. In these situations, we suggest to approximate the prior by using a generalized beta prime distribution that matches it closely. This approach is quite flexible and can be easily implemented in standard Bayesian software. Lastly, we demonstrate the performance of the method on simulated data where it particularly shines in high-dimensional examples as well as real-world data which shows its ability to model spatial correlation in the random effects.
Many real-world networks are theorized to have core-periphery structure consisting of a densely-connected core and a loosely-connected periphery. While this network feature has been extensively studied in various scientific disciplines, it has not received sufficient attention in the statistics community. In this expository article, our goal is to raise awareness about this topic and encourage statisticians to address the many interesting open problems in this area. We present the current research landscape by reviewing the most popular metrics and models that have been used for quantitative studies on core-periphery structure. Next, we formulate and explore various inferential problems in this context, such as estimation, hypothesis testing, and Bayesian inference, and discuss related computational techniques. We also outline the multidisciplinary scientific impact of core-periphery structure in a number of real-world networks. Throughout the article, we provide our own interpretation of the literature from a statistical perspective, with the goal of prioritizing open problems where contribution from the statistics community will be effective and important.
Networks continue to be of great interest to statisticians, with an emphasis on community detection. Less work, however, has addressed this question: given some network, does it exhibit meaningful community structure? We propose to answer this question in a principled manner by framing it as a statistical hypothesis in terms of a formal and model-agnostic homophily metric. Homophily is a well-studied network property where intra-community edges are more likely than between-community edges. We use the homophily metric to identify and distinguish between three concepts: nominal, collateral, and intrinsic homophily. We propose a simple and interpretable test statistic leveraging this homophily parameter and formulate both asymptotic and bootstrap-based rejection thresholds. We prove its asymptotic properties and demonstrate it outperforms benchmark methods on both simulated and real world data. Furthermore, the proposed method yields rich, provocative insights on classic data sets; namely, that meany well-studied networks do not actually have intrinsic homophily.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.