R/faConfInt.R

In userfriendlyscience: Quantitative Analysis Made Accessible

#' Extract confidence bounds from psych's factor analysis object
#' 
#' This function contains some code from a function in \code{\link{psych}}
#' that's not exported \code{print.psych.fa.ci} but useful nonetheless. It
#' basically takes the outcomes of a factor analysis and extracted the
#' confidence intervals.
#' 
#' THis function extract confidence interval bounds and combines them with
#' factor loadings using the code from the \code{print.psych.fa.ci} in
#' \code{\link{psych}}.
#' 
#' @param fa The object produced by the \code{\link{fa}} function from the
#' \code{\link{psych}} package. It is important that the \code{n.iter} argument
#' of \code{\link{fa}} was set to a realistic number, because otherwise, no
#' confidence intervals will be available.
#' @return A list of dataframes, one for each extracted factor, with in each
#' dataframe three variables: \item{lo}{lower bound of the confidence interval}
#' \item{est}{point estimate of the factor loading} \item{hi}{upper bound of
#' the confidence interval}
#' @author William Revelle (extracted by Gjalt-Jorn Peters)
#' 
#' Maintainer: Gjalt-Jorn Peters <gjalt-jorn@@userfriendlyscience.com>
#' @keywords ~kwd1 ~kwd2
#' @examples
#' 
#' \dontrun{
#' ### Not run because it takes too long to run to test it,
#' ### and may produce warnings, both because of the bootstrapping
#' ### required to generate the confidence intervals in fa
#' faConfInt(fa(Thurstone.33, 2, n.iter=100, n.obs=100));
#' }
#' 
#' @export faConfInt
faConfInt <- function(fa) {

  ### Combine both confidence intervals and factor loadings, using
  ### the code from the 'psych:::print.psych.fa.ci' function 
  lc <- data.frame(unclass(fa$loadings), fa$ci$ci);
  ### Create list for CIs per factor
  CIs <- list();
  for (i in 1:fa$factors) {
    CIs[[i]] <- lc[, c(i + fa$factors, i, i + fa$factors * 2)];
    names(CIs[[i]]) <- c('lo', 'est', 'hi');
  }
  
  return(CIs);
  
}