Autoregressive (AR) models are some of the most widely used models in applied economics, among other disciplines, because of their generality and simplicity. However, the dynamic characteristics of real economic and financial data can change from one time period to another, limiting the applicability of linear time-series models. For example, the change of unemployment rate is a function of the state of the economy, whether it is expanding or contracting. A variety of models have been developed that allow time-series dynamics to depend on the regime of the system they are part of. The class of regime-dependent models include Markov-switching, smooth transition, and threshold autoregressive (TAR) models. Read more…

Monitoring lineages of the Omicron variant of the SARS-CoV-2 virus continues to be an important health consideration. The World Health Organization identifies BA.1, BA.1.1, and the most recent BA.2 as the most common sublineages. A recent study from Japan, Yamasoba et al. (2022), compares, among other characteristics, the transmissibility of these three Omicron lineages with the latest Delta variant. It identifies BA.2 to have the highest transmissibility of the four. Preprint of the study is available at bioarxiv.org. One interesting aspect of the study is the application of Bayesian multilevel models for representing lineage growth dynamics. In this post, I demonstrate how to use Stata’s bayesmh and bayesstats summary commands to perform similar analysis. Read more…

Overview

Markov chain Monte Carlo (MCMC) is the principal tool for performing Bayesian inference. MCMC is a stochastic procedure that utilizes Markov chains simulated from the posterior distribution of model parameters to compute posterior summaries and make predictions. Given its stochastic nature and dependence on initial values, verifying Markov chain convergence can be difficult—visual inspection of the trace and autocorrelation plots are often used. A more formal method for checking convergence relies on simulating and comparing results from multiple Markov chains; see, for example, Gelman and Rubin (1992) and Gelman et al. (2013). Using multiple chains, rather than a single chain, makes diagnosing convergence easier.

As of Stata 16, bayesmh and its bayes prefix commands support a new option, nchains(), for simulating multiple Markov chains. There is also a new convergence diagnostic command, bayesstats grubin. All Bayesian postestimation commands now support multiple chains. In this blog post, I show you how to check MCMC convergence and improve your Bayesian inference using multiple chains through a series of examples. I also show you how to speed up your sampling by running multiple Markov chains in parallel. Read more…

Introduction

Stata 15 provides a convenient and elegant way of fitting Bayesian regression models by simply prefixing the estimation command with bayes. You can choose from 45 supported estimation commands. All of Stata’s existing Bayesian features are supported by the new bayes prefix. You can use default priors for model parameters or select from many prior distributions. I will demonstrate the use of the bayes prefix for fitting a Bayesian logistic regression model and explore the use of Cauchy priors (available as of the update on July 20, 2017) for regression coefficients. Read more…

As of Stata 16, see [BAYES] bayesstats grubin and Bayesian analysis: Gelman-Rubin convergence diagnostic.

The original blog posted May 26, 2016, omitted option initrandom from the bayesmh command. The code and the text of the blog entry were updated on August 9, 2018, to reflect this.

Overview

MCMC algorithms used for simulating posterior distributions are indispensable tools in Bayesian analysis. A major consideration in MCMC simulations is that of convergence. Has the simulated Markov chain fully explored the target posterior distribution so far, or do we need longer simulations? A common approach in assessing MCMC convergence is based on running and analyzing the difference between multiple chains.

For a given Bayesian model, bayesmh is capable of producing multiple Markov chains with randomly dispersed initial values by using the initrandom option, available as of the update on 19 May 2016. In this post, I demonstrate the Gelman–Rubin diagnostic as a more formal test for convergence using multiple chains. For graphical diagnostics, see Graphical diagnostics using multiple chains in [BAYES] bayesmh for more details. To compute the Gelman–Rubin diagnostic, I use an unofficial command, grubin, which can be installed by typing the following in Stata: Read more…

This post was written jointly with Yulia Marchenko, Executive Director of Statistics, StataCorp.

As of update 03 Mar 2016, bayesmh provides a more convenient way of fitting distributions to the outcome variable. By design, bayesmh is a regression command, which models the mean of the outcome distribution as a function of predictors. There are cases when we do not have any predictors and want to model the outcome distribution directly. For example, we may want to fit a Poisson distribution or a binomial distribution to our outcome. This can now be done by specifying one of the four new distributions supported by bayesmh in the likelihood() option: dexponential(), dbernoulli(), dbinomial(), or dpoisson(). Previously, the suboption noglmtransform of bayesmh‘s option likelihood() was used to fit the exponential, binomial, and Poisson distributions to the outcome variable. This suboption continues to work but is now undocumented.

For examples, see Beta-binomial model, Bayesian analysis of change-point problem, and Item response theory under Remarks and examples in [BAYES] bayesmh.

We have also updated our earlier “Bayesian binary item response theory models using bayesmh” blog entry to use the new dbernoulli() specification when fitting 3PL, 4PL, and 5PL IRT models.

This post was written jointly with Yulia Marchenko, Executive Director of Statistics, StataCorp.

Overview
1PL model
2PL model
3PL model
4PL model
5PL model
Conclusion

Item response theory (IRT) is used for modeling the relationship between the latent abilities of a group of subjects and the examination items used for measuring their abilities. Stata 14 introduced a suite of commands for fitting IRT models using maximum likelihood; see, for example, the blog post Spotlight on irt by Rafal Raciborski and the [IRT] Item Response Theory manual for more details. In this post, we demonstrate how to fit Bayesian binary IRT models by using the redefine() option introduced for the bayesmh command in Stata 14.1. We also use the likelihood option dbernoulli() available as of the update on 03 Mar 2016 for fitting Bernoulli distribution. If you are not familiar with the concepts and jargon of Bayesian statistics, you may want to watch the introductory videos on the Stata Youtube channel before proceeding.

Introduction to Bayesian analysis, part 1 : The basic concepts
Introduction to Bayesian analysis, part 2: MCMC and the Metropolis-Hastings algorithm

We use the abridged version of the mathematics and science data from DeBoeck and Wilson (2004), masc1. The dataset includes 800 student responses to 9 test questions intended to measure mathematical ability.

The irt suite fits IRT models using data in the wide form – one observation per subject with items recorded in separate variables. To fit IRT models using bayesmh, we need data in the long form, where items are recorded as multiple observations per subject. We thus reshape the dataset in a long form: we have a single binary response variable, y, and two index variables, item and id, which identify the items and subjects, respectively. This allows us to Read more…