Generate an empirical objective function from default or empirical values for use in an item selection using STUART.
empiricalobjective(
criteria = c("rmsea", "srmr", "crel"),
add = c("chisq", "df", "pvalue"),
x = NULL,
n = 50,
side = NULL,
skew = FALSE,
scale = 1,
matrices = NULL,
fixed = NULL,
comparisons = NULL,
...
)Returns an object of class stuartFixedObjective
A vector of names of criteria included in the objective function. Defaults to c('rmsea', 'srmr', 'crel').
A vector of names of criteria not used in the objective function, but added in order to be included in the log of solutions.
Either a vector of values or an object of class stuartOutput from which to determine values in the objective function. If NULL (the default) values are generated from criteria-specific presets.
Number of values to use in function determining. Defaults to 50, meaning if side = 'top' the 50 largest values are used to determine discrimination and difficulty parameters for each criterion.
Which side good values are located at. 'top' means large values are good (e.g. Reliability), 'bottom' means small values are good (e.g. RMSEA), and 'middle' means average values are good (e.g. factor correlations).
Whether to account for skew in the distribution using the psn() function from the sn-Package. Defaults to FALSE, meaning a normal distribution is used.
A numeric scale to use in weighting the objective component. Defaults to 1.
An object of class stuartObjectiveMatrices to include matrices (e.g. latent correlations) into the objective function.
An object of class stuartFixedObjective to include already previously defined fixed objectives.
A vector of names of criteria included in the objective function which are related to model comparisons (e.g. when determining measurement invariance).
Additional arguments.
Martin Schultze
fixedobjective, extractobjective, objectivematrices