1. What are the contributions in "Bayesian hierarchical classes analysis" ?
In this paper, the authors present a stochastic extension of the hierarchical classes model for two-way two-mode binary data.. A fully Bayesian method for fitting the new model is presented and evaluated in a simulation study.. Furthermore, the authors propose tools for model selection and model checking based on Bayes factors and posterior predictive checks.
read more
2. What are the future works in "Bayesian hierarchical classes analysis" ?
Finally, the new model extension as presented here may be extended further in various directions.. Finally, an interesting direction for further research is charting out by means of a sensitivity analysis how and to which extent the obtained simulated posterior distribution may depend on the selection of the prior.. A second direction in which the model can be further extended relates to the choice of the prior distribution.
read more
3. How many times did the confidence interval contain the true value of the data?
Given that the Bayesian confidence interval almost always (i.e., in 99.9% of the cases) contains the true value, the authors can derive that a Bayesian analysis succeeds in recovering the true value of about 90% of the cells in S and P with virtual certainty.
read more
4. What is the way to assess the goodness of recovery of a bundle value parameter?
With respect to the goodness of recovery of a bundle value parameter sik or pjk, remind that the true value is either 0 or 1 and that also the Bayesian confidence interval has a width of either 0 (i.e., the 2.5 and 97.5 percentiles have the same value) or 1 (i.e., the 2.5 and 97.5 percentiles equal 0 and 1, respectively).
read more
