Plot Method for the Output of 'q_mediation' Family — plot.q

Plot the path model fitted by the family of 'q_mediation' functions.

Usage

# S3 method for class 'q_mediation'
plot(
  x,
  standardized = FALSE,
  size_variables = NULL,
  size_path_labels = NULL,
  nchar_variables = NULL,
  nchar_path_labels = NULL,
  digits = 2,
  rsquares = TRUE,
  sigs = TRUE,
  margins = c(5, 5, 5, 5),
  v_pos = c("middle", "lower", "upper"),
  v_preference = c("upper", "lower"),
  print_indirect = TRUE,
  indirect_standardized = c("none", "stdx", "stdy", "stdxy"),
  size_indirect = 1,
  plot_now = TRUE,
  ...
)

indirect_on_plot(
  q_mediation_output = NULL,
  digits = 2,
  size_indirect = 1,
  indirect_standardized = c("none", "stdx", "stdy", "stdxy"),
  margins = c(5, 5, 5, 5),
  original_plot = NULL
)

Arguments

x: The output of q_mediation(), q_simple_mediation(), q_serial_mediation(), and q_parallel_mediation(). (Named x because it is required in the naming of arguments of the plot generic function.)
standardized: Logical. If TRUE, betaS in the printout of q_mediation() family will be used in the figure, with only numerical variables standardized. If FALSE, the default, then the original (unstandardized) coefficients will be used.
size_variables: The size of the observed variables (the "rectangles"), to be passed to sizeMan of semPlot::semPaths(). Default is NULL and the size is determined internally based on the number of variables.
size_path_labels: The size of the edge labels (parameter estimates), to be passed to edge.label.cex of semPlot::semPaths(). Default is NULL and the size is determined internally based on the number of variables in the plot.
nchar_variables: The number of characters to be displayed for each variable. To be passed to nCharNodes of semPlot::semPaths(). Default is NULL, equivalent to 0 for nCharNodes, to disable abbreviating the variable names.
nchar_path_labels: The number of characters to be displayed for each label for a path. To be passed to nCharEdges of semPlot::semPaths(). Default is NULL, equivalent to 0 for nCharEdges, to disable abbreviating the labels.
digits: The number of digits to be printed after the decimals. To be passed to nDigits of semPlot::semPaths(). Default is 2.
rsquares: Logical. If TRUE, the default, R-squares will be drawn instead of error variances for mediators and outcome variables (the y variables).
sigs: Logical. If TRUE, the default, significance test results will be marked by asterisks, based on the same p-values for R-squares displayed when printing the output of the q_mediation() family.
margins: The margins of the plot. A numeric vector of four values: bottom, left, top, and right. Passed to the mar argument of semPlot::semPaths().
v_pos: How the mediators are to be positioned vertically. If set to "middle", with one x variable and one y variable, the mediators will tend to be placed around the horizontal line joining x and y. If set to "upper", they will be placed along or above this line. If set to "lower", they will be placed along or below this line. Note that this only affects the initial positions. The positions will be further adjusted based on the free paths in the model. This argument is to be passed to semptools::auto_layout_mediation().
v_preference: The preference in shifting the mediators upward ("upper") or downward ("lower") to avoid blocking or overlapping with any paths in the models. It is used only whenv_pos is "middle". If v_pos is "lower", then v_preference will be forced to be "lower". If v_pos is "upper", then v_preference will be forced to be "upper". This argument is to be passed to semptools::auto_layout_mediation().
print_indirect: Logical. Whether the indirect effect(s), and total indirect effect if applicable, will be printed on the plot. Default is TRUE. Used only if plot_now is TRUE. Confidence intervals, if stored, will be printed, at the level of confidence used when doing the analysis.
indirect_standardized: If print_indirect is TRUE, which type of effects are to be printed: "none" for the unstandardized (raw) indirect effects, "stdx" for the effects with x standardized, "stdy" for the effects with y standardized, and "stdxy" for the effects with both x and y standardized.
size_indirect: The size used when printing the indirect effects. The final size is determined by multiplying the final value size_path_labels (determined internally if it is set to NULL) by this value. If equal to 1, then the size used in printing the indirect effects should be "close" to the size of numbers on the paths.
plot_now: If TRUE, the default, the plot will be plotted when calling this method.
...: For the plot method, these are optional arguments to be passed to semPlot::semPaths() to generate the initial plot, before being adjusted by semptools::auto_layout_mediation().
q_mediation_output: The original object used to generate the plot (the output of the q_mediation() family). Indirect effects will be retrieved from this output.
original_plot: The plot generated by the plot method. If supplied, a new plot will be generated and then the indirect effects will be printed on this new plot. If NULL, the default, the indirect effects will be printed on the existing plot. Space will be added to make room for the indirect effects only if this argument is set. If original_plot is not used, make sure there is enough room at the bottom for the indirect effects.

Value

The plot method returns a qgraph object generated by semPlot::semPaths(), which is plotted by default unless plot_now is set to FALSE. It can be further modified by other functions that work on a qgraph object, such as those from semptools.

The function indirect_on_plot() returns the object set to q_mediation_output invisibly. It is called for its side-effect.

Details

This method requires the semptools and semPlot packages. They are not installed by default. Install them first before using the plot method.

This method draws the path models fitted by q_mediation(), q_simple_mediation(), q_serial_mediation(), and q_parallel_mediation(), with path coefficients and R-squares.

It will try to set positions of the variables automatically, following the left-to-right convention: x variables on the left, y variables on the right, mediators between them, and arrows (paths) flow from left to right. The figure should usually be usable. If not, it can be further modified by helper functions such as those in semptools that can manipulate a qgraph object. For example, semptools::move_node() can be used to adjust the position of a variable in the plot.

The helper function indirect_on_plot() adds the indirect effect estimates (as well as confidence intervals and p-values, if available) to a plot. The plot method will add these effects by default, and so users usually do not need to use this function. However, if the plot needs to be modified before being drawn, this function can be used to add the effects after drawing the modified plot.

Examples


# These examples require the package
# semptools (version 0.3.2 or above).

# CI disabled in these examples.
# Please see the help page of these functions on forming
# confidence intervals for the indirect effects.

# ===== Simple mediation

out <- q_simple_mediation(x = "x",
                          y = "y",
                          m = "m",
                          cov = c("c2", "c1"),
                          boot_ci = FALSE,
                          data = data_med)
#> - Compute unstandardized indirect effect(s) ....
#> - Compute standardized-y indirect effect(s) ....
#> - Compute standardized-x indirect effect(s) ....
#> - Compute standardized-x-and-y indirect effect(s) ....
#> - Compute total indirect effect(s) ....
#> - Compute the direct effect ....
#> - Compute the standardized-y direct effect ....
#> - Compute the standardized-x direct effect ....
#> - Compute the standardized-x-and-y direct effect ....
#> Computation completed.
plot(out)


# ===== Serial mediation

out <- q_serial_mediation(x = "x",
                          y = "y",
                          m = c("m1", "m2"),
                          cov = c("c2", "c1"),
                          boot_ci = FALSE,
                          data = data_serial)
#> - Compute unstandardized indirect effect(s) ....
#> - Compute standardized-y indirect effect(s) ....
#> - Compute standardized-x indirect effect(s) ....
#> - Compute standardized-x-and-y indirect effect(s) ....
#> - Compute total indirect effect(s) ....
#> - Compute the direct effect ....
#> - Compute the standardized-y direct effect ....
#> - Compute the standardized-x direct effect ....
#> - Compute the standardized-x-and-y direct effect ....
#> Computation completed.
plot(out)

# Standardized effects
plot(out,
     standardized = TRUE,
     indirect_standardized = "stdxy")


# ==== Parallel mediation

out <- q_parallel_mediation(x = "x",
                            y = "y",
                            m = c("m1", "m2"),
                            cov = c("c2", "c1"),
                            boot_ci = FALSE,
                            data = data_parallel)
#> - Compute unstandardized indirect effect(s) ....
#> - Compute standardized-y indirect effect(s) ....
#> - Compute standardized-x indirect effect(s) ....
#> - Compute standardized-x-and-y indirect effect(s) ....
#> - Compute total indirect effect(s) ....
#> - Compute the direct effect ....
#> - Compute the standardized-y direct effect ....
#> - Compute the standardized-x direct effect ....
#> - Compute the standardized-x-and-y direct effect ....
#> Computation completed.

plot(out)

plot(out,
     v_pos = "lower")

plot(out,
     v_pos = "upper")


# ===== A user-specified mediation model

out <- q_mediation(x = "x1",
                   y = "y1",
                   model = c("x1 -> m11 -> m2 -> y1",
                             "x1 -> m12 -> m2 -> y1"),
                   cov = c("c2", "c1"),
                   boot_ci = FALSE,
                   data = data_med_complicated)
#> - Compute unstandardized indirect effect(s) ....
#> - Compute standardized-y indirect effect(s) ....
#> - Compute standardized-x indirect effect(s) ....
#> - Compute standardized-x-and-y indirect effect(s) ....
#> - Compute total indirect effect(s) ....
#> - No direct path from x1 to y1 in the model. Skip the computation of direct effect ...
#> Computation completed.
plot(out)