calibrateItems: Determine the optimal scale constant for a set of items

Description

Data are passed through filterGraph and normalizeData. Then the ‘unidim_adapt’ model is fit to each item individually. A larger varCorrection will obtain a more accurate scale, but is also more likely to produce an intractable model. A good compromise is between 5.0 and 9.0.

Usage

calibrateItems(
  df,
  iter = 2000L,
  chains = 4L,
  varCorrection = 5,
  maxAttempts = 5L,
  ...
)

Value

A data.frame (one row per item) with the following columns:

item: Name of the item
iter: Number of iterations per chain
divergent: Number of divergent transitions observed after warmup
treedepth: Number of times the treedepth was exceeded
low_bfmi: Number of chains with low E-BFMI
n_eff: Minimum effective number of samples across all parameters
Rhat: Maximum Rhat across all parameters
scale: Median marginal posterior of scale
thetaVar: Median variance of theta (latent scores)

Arguments

df: a data frame with pairs of vertices given in columns pa1 and pa2, and item response data in other columns
iter: A positive integer specifying the number of iterations for each chain (including warmup).
chains: A positive integer specifying the number of Markov chains.
varCorrection: A correction factor greater than or equal to 1.0
maxAttempts: How many times to try re-running a model with more iterations.
...: Additional options passed to stan.

References

Vehtari, A., Gelman, A., Simpson, D., Carpenter, B., & Bürkner, P. C. (2019). Rank-normalization, folding, and localization: An improved \(\widehat R\) for assessing convergence of MCMC. arXiv preprint arXiv:1903.08008.

Examples

Run this code

# \donttest{
result <- calibrateItems(phyActFlowPropensity)  # takes more than 5 seconds
print(result)
# }