High-level validation results

Coverage of quantiles from 1,000 analyses of datasets simulated under the prior predictive.

data(sbc_quants)

Format

data.frame with nominal quantiles and observed coverage.

quantile

The quantile for which the CI was computed.

r_covered

The proportion of true values of R covered at the quantile.

r_covered

The proportion of true values of k covered at the quantile.

Source

Generated in R using nbbp.

Details

When datasets are simulated from the prior predictive, frequentist coverage properties of the posterior quantiles should hold. This dataframe contains the results from analyzing 1,000 datasets simulated from the prior predictive (under default priors) of 25 observations each. Thus, a plot of quantiles and coverage at those quantiles should fall near the 1:1 line.

Only analyses meeting convergence standards (min(n_eff) > 1000, max(Rhat) < 1.005) are included in those used to compute coverage.

Examples

if (FALSE) { # \dontrun{
data(sbc_quants)
plot(sbc_quants$quantile, sbc_quants$r_covered)
abline(a = 0, b = 1)
} # }