Table 1.
Discrepancy functions considered in the simulations of this paper.
Table 2.
Summary of the models considered in simulations.
Figure 1.
Power of the sampled posterior (; solid line), half-sample ML (
; dotted line) and posterior predictive (
; dotted-dashed line) p-values.
Power of the p-values (solid line),
(dotted line) and
(dotted-dashed line) used with the maximum test statistic to detect departures from the Poisson distribution at the level of p = 0.05 when data are distributed according to a Polya distribution with
. Power is plotted as a function of sample size N varying between 20 and 80.
and
were equivalent in terms of power, and both were more powerful than
except at the highest sample sizes. The dotted baseline level corresponds to p = 0.05.
Table 3.
Behavior of relative to the uniform distribution under Scenario 2, depending on the interval housing the statistical prior sharpness parameter
.
Table 4.
Behavior of relative to the uniform distribution under Scenario 3, based on the frequency of
values found at the 5% extremities of the unit interval, depending on the interval of the statistical prior sharpness parameter
(in rows) and off-centering parameter
(in columns).
Table 5.
Difference in power between or
and
according to sample size (in columns) and discrepancy function (in rows).