Bayesian unanchored additive models for component network meta-analysis.

Component network meta-analysis (CNMA) models are an extension of standard network meta-analysis (NMA) models which account for the use of multicomponent treatments in the network. This article contributes innovatively to several statistical aspects of CNMA. First, by introducing a unified notation, we establish that currently available methods differ in the way they assume additivity, an important distinction that has been overlooked so far in the literature. In particular, one model uses a more restrictive form of additivity than the other which we term an anchored and unanchored model, respectively. We show that an anchored model can provide a poor fit to the data if it is misspecified. Second, given that Bayesian models are often preferred by practitioners, we develop two novel unanchored Bayesian CNMA models presented under the unified notation. An extensive simulation study examining bias, coverage probabilities, and treatment rankings confirms the favorable performance of the novel models. This is the first simulation study to compare the statistical properties of CNMA models in the literature. Finally, the use of our novel models is demonstrated on a real dataset, and the results of CNMA models on the dataset are compared.

Estimation and Applications of Uncertainty in Methane Emissions Quantification Technologies: A Bayesian Approach.

An accurate understanding of uncertainty is needed to properly interpret methane emission estimates from upstream oil and gas sources in a variety of contexts, from component-level measurements to yearly jurisdiction-wide inventories. To characterize measurement uncertainty, we examine controlled release (CR) data from five different technology providers including quantitative gas imaging (QOGI), tunable diode laser-absorption spectroscopy (TDLAS); and airborne near-infrared hyperspectral (NIR HS) imaging. We introduce a novel empirical method to develop probability distributions of measurements given a true emission rate using the CR data. The approach includes flexible likelihoods which capture complex relationships in the data. An algorithm which provides the distribution of the true emission rate given a measurement is also developed, which synthesizes the measurement with the CR data and external information about the possible true emission rate. The results show that flexible models that accommodate complex nonlinear behavior are needed to adequately model measurement error. We also show that measurement error can vary under different conditions. We demonstrate that measurement uncertainty can be reduced by performing repeated measurements. A limitation of the study is that the collected CR data is collected under controlled conditions that may differ from those in industrial settings. As new CR data become available, the models presented in this paper can be refit to consider more diverse scenarios. The methodology can be extended to explicitly model different conditions to improve performance.