Currently Browsing: Tools

Bayes Factors for Stan Models without Tears

For Christian Robert’s blog post about the bridgesampling package, click here.

Bayesian inference is conceptually straightforward: we start with prior uncertainty and then use Bayes’ rule to learn from data and update our beliefs. The result of this learning process is known as posterior uncertainty. Quantities of interest can be parameters (e.g., effect size) within a single statistical model or different competing models (e.g., a regression model with three predictors vs. a regression model with four predictors). When the focus is on models, a convenient way of comparing two models M1 and M2 is to consider the model odds:


    \begin{equation*} \label{eq:post_model_odds} \underbrace{\frac{p(\mathcal{M}_1 \mid \text{data})}{p(\mathcal{M}_2 \mid \text{data})}}_{\text{posterior odds}} = \underbrace{\frac{p(\text{data} \mid \mathcal{M}_1)}{p(\text{data} \mid \mathcal{M}_2)}}_{\text{Bayes factor BF$_{12}$}} \times \underbrace{\frac{p(\mathcal{M}_1)}{p(\mathcal{M}_2)}}_{\text{prior odds}}. \end{equation*}



An Interactive App for Designing Informative Experiments

Bayesian inference offers the pragmatic researcher a series of perks (Wagenmakers, Morey, & Lee, 2016). For instance, Bayesian hypothesis tests can quantify support in favor of a null hypothesis, and they allow researchers to track evidence as data accumulate (e.g., Rouder, 2014).

However, Bayesian inference also confronts researchers with new challenges, for instance concerning the planning of experiments. Within the Bayesian paradigm, is there a procedure that resembles a frequentist power analysis? (yes, there is!)


Powered by WordPress | Designed by Elegant Themes