Skip to contents

This function creates a plot of the estimates for covariate coefficients, obtained from the first step (univariate testing) in each scm performed in the bootscm. When normalized for their standard deviation, these plots can be used to compare the strength of the covariate relationship. Coloring is based on the covariate being included in the final model (blue) not being included (red).

Usage

xp.boot.par.est(
  bootgam.obj = NULL,
  sd.norm = TRUE,
  by.cov.type = FALSE,
  abs.values = FALSE,
  show.data = TRUE,
  show.means = TRUE,
  show.bias = TRUE,
  dotpch = c(1, 19),
  labels = NULL,
  pch.mean = "|",
  xlab = NULL,
  ylab = NULL,
  col = c(rgb(0.8, 0.5, 0.5), rgb(0.2, 0.2, 0.7), rgb(0.2, 0.2, 0.7), rgb(0.6, 0.6, 0.6)),
  ...
)

Arguments

bootgam.obj

The object created using bootscm.import(), which hold the data for plotting.

sd.norm

Perform normalization of the covariate coefficients (default is TRUE). When TRUE, the estimated covariate coefficients will be multiplied by the standard deviation of the specific covariate (both for continuous and categorical covariates).

by.cov.type

Split the plot for continuous and dichotomous covariates. Default is FALSE.

abs.values

Show the covariate coefficient in absolute values. Default is FALSE.

show.data

Show the actual covariate coefficients in the plot. Default is TRUE.

show.means

Show the means of included covariates (blue) and all covariates (grey) in the plot. Default is TRUE.

show.bias

Show estimated bias as text in the plot. Default is TRUE.

dotpch

The character used for plotting.

labels

Custom labels for the parameter-covariate relationships, (character vector)

pch.mean

The character used for plotting the mean.

xlab

Custom x-axis label

ylab

Custom y-axis label

col

The color scheme.

...

Additional plotting arguments may be passed to this function.

Value

No value returned.

Details

Optionally, estimated bias is plotted in the graph (as text). Bias is also shown by the difference in mean of parameter estimates when the covariate is included (blue diamond), as opposed to the mean of all parameter estimates (grey diamond)

Note: For dichotomous covariates, the default PsN implementation is to use the most common covariate value as base, while the effect of the other value, is estimated by a theta. Xpose (bootscm.import) however recalculates the estimated parameters, to the parametrization in which the lowest value of the dichotomous covariate is the base (e.g. 0), and the estimated THETA denotes the proportional change, when the covariate has the other value (e.g. 1).

Author

Ron Keizer

Examples


  xp.boot.par.est()
#> boot.type bootgam.objData not available. Did you import the bootSCM data?
#> NULL