semptools (version 0.3.2)

mark_se: Add Standard Error Estimates to Parameter Estimates (Edge Labels)

Description

Add standard error estimates, in parentheses, to parameter estimates (edge labels) in a qgraph::qgraph object.

Usage

mark_se(
  semPaths_plot,
  object,
  sep = " ",
  digits = 2L,
  ests = NULL,
  std_type = FALSE
)

Value

If the input is a qgraph::qgraph object, the function returns a qgraph based on the original one, with standard error estimates appended. If the input is a list of qgraph objects, the function returns a list of the same length.

Arguments

semPaths_plot

A qgraph object generated by semPaths, or a similar qgraph object modified by other semptools functions.

object

The object used by semPaths to generate the plot. Use the same argument name used in semPaths to make the meaning of this argument obvious. Currently only object of class lavaan is supported.

sep

A character string to separate the coefficient and the standard error (in parentheses). Default to " " (one space). Use "\n" to enforce a line break.

digits

Integer indicating number of decimal places for the appended standard errors. Default is 2L.

ests

A data.frame from the parameterEstimates function, or from other function with these columns:? lhs, op, rhs, and se. Only used when object is not specified.

std_type

If standardized solution is used in the plot, set this either to the type of standardization (e.g., "std.all") or to TRUE. It will be passed to lavaan::standardizedSolution() to compute the standard errors for the standardized solution. Used only if standard errors are not supplied directly through ests.

Details

Modify a qgraph::qgraph object generated by semPaths (currently in parentheses) to the labels. Require either the original object used in the semPaths call, or a data frame with the standard error for each parameter. The latter option is for standard errors not computed by lavaan but by other functions.

Currently supports only plots based on lavaan output.

This function is a variant of, and can be combined with, the mark_sig function.

Examples

mod_pa <-
  'x1 ~~ x2
   x3 ~  x1 + x2
   x4 ~  x1 + x3
  '
fit_pa <- lavaan::sem(mod_pa, pa_example)
lavaan::parameterEstimates(fit_pa)[ , c("lhs", "op", "rhs",
                                       "est", "pvalue", "se")]
m <- matrix(c("x1",   NA,   NA,
               NA, "x3", "x4",
             "x2",   NA,   NA), byrow = TRUE, 3, 3)
p_pa <- semPlot::semPaths(fit_pa, whatLabels = "est",
                         style = "ram",
                         nCharNodes = 0, nCharEdges = 0,
                         layout = m)
p_pa2 <- mark_se(p_pa, fit_pa)
plot(p_pa2)

mod_cfa <-
 'f1 =~ x01 + x02 + x03
  f2 =~ x04 + x05 + x06 + x07
  f3 =~ x08 + x09 + x10
  f4 =~ x11 + x12 + x13 + x14
 '
fit_cfa <- lavaan::sem(mod_cfa, cfa_example)
lavaan::parameterEstimates(fit_cfa)[ , c("lhs", "op", "rhs",
                                        "est", "pvalue", "se")]
p_cfa <- semPlot::semPaths(fit_cfa, whatLabels = "est",
                          style = "ram",
                          nCharNodes = 0, nCharEdges = 0)
# Place standard errors on a new line
p_cfa2 <- mark_se(p_cfa, fit_cfa, sep = "\n")
plot(p_cfa2)

mod_sem <-
'f1 =~ x01 + x02 + x03
 f2 =~ x04 + x05 + x06 + x07
 f3 =~ x08 + x09 + x10
 f4 =~ x11 + x12 + x13 + x14
 f3 ~  f1 + f2
 f4 ~  f1 + f3
'
fit_sem <- lavaan::sem(mod_sem, sem_example)
lavaan::parameterEstimates(fit_sem)[ , c("lhs", "op", "rhs",
                                        "est", "pvalue", "se")]
p_sem <- semPlot::semPaths(fit_sem, whatLabels = "est",
                          style = "ram",
                          nCharNodes = 0, nCharEdges = 0)
# Mark significance, and then add standard errors
p_sem2 <- mark_sig(p_sem, fit_sem)
p_sem3 <- mark_se(p_sem2, fit_sem, sep = "\n")
plot(p_sem3)