get_default_jags_model: Write a default jags model

In snsf-data/ERforResearch: Expected Rank for Research Evaluation

Write a default jags model

Description

This function writes a default JAGS model, given the parameters and stores it as a. txt in a certain path.

Usage

get_default_jags_model(
  outcome_variable = "continuous",
  nine_point_scale_continuous = FALSE,
  residuals = "homogeneous",
  subpanels = FALSE,
  path = "default_jags_model.txt",
  quiet = FALSE
)

Arguments

outcome_variable	the nature of the outcome variable: continuous or ordinal. By default the grades are assumed to be '"continuous"'.
nine_point_scale_continuous	boolean variable indicated whether or not a nine-point continuous scale is used. By default is is set to 'FALSE', which means that a six-point scale is used by default. Be aware that those specifications are due to the usage of the package by the Swiss National Science Foundation.
residuals	the nature of the residuals in the model: homogeneous or heterogeneous. By default the residuals are assumed to be '"homogeneous"'.
subpanels	the individual votes come from different sub-panels that need to be merged. By default this is set to 'FALSE' meaning that rankings are produced for each panel separately.
path	where should it be stored, relative to getwd()? By default this is set to '"jags_model.txt"'. Hence it is simply written to the working directory.
quiet	if set to 'TRUE', the function does not show any messages of caution.

Details

Note that many of the specific default parameters are specified for easy usage and integration in the funding evaluation system at the Swiss National Science Foundation and might not be applicable to other evaluation systems.

The model defined here has a random component for the proposal and the assessor. If the parameter subpanels is set to TRUE an additional grouping for the panel is defined. The user is invited to write their own model definition if more flexibility is needed.

The path to a user-specified model can then be given as parameter to the 'get_mcmc_samples()' or the 'get_er_from_jags()' functions. The user can also simply decide between a continuous or ordinal outcome variable, homogeneous or heterogeneous residuals, with or without subpanel merging (those options are all integrated), write the model in a .txt file and adapt it from there.

Examples

# The model definition .txt is stored in the file "default_jags_model.txt"
## Not run: 
get_default_jags_model()

## End(Not run)

snsf-data/ERforResearch documentation built on May 12, 2024, 9:25 p.m.