Mediation — pamlmed • pamlj

Power analysis for simple and multiple mediation models using the Sobel test, joint-significance, or bootstrap confidence intervals.

pamlmed(
  aim = "n",
  mode = "medsimple",
  a = 0.3,
  b = 0.3,
  cprime = 0,
  model_type = "twomeds",
  a1 = "",
  b1 = "",
  a2 = "",
  b2 = "",
  a3 = "",
  b3 = "",
  d1 = "",
  d2 = "",
  r12 = "",
  r13 = "",
  r23 = "",
  cprime2 = 0,
  sensitivity_coef = "a1",
  power = 0.9,
  n = 100,
  sig.level = 0.05,
  alternative = "two.sided",
  test = "joint",
  mcR = 1000,
  parallel = FALSE,
  set_seed = FALSE,
  seed = 42,
  table_pwbyn = TRUE,
  table_pwbyes = FALSE,
  plot_ncurve = FALSE,
  plot_log = FALSE,
  plot_palette = "viridis",
  plot_x = "none",
  plot_y = "none",
  plot_custom_labels = FALSE,
  plot_z = "none",
  plot_x_from = 0,
  plot_x_to = 0,
  plot_z_lines = 1,
  plot_z_value = list(),
  plot_to_table = FALSE,
  test_c = FALSE,
  explain = FALSE,
  diagram = TRUE
)

Arguments

aim: The aim of the analysis: n (default) for sample size, power to estimate power, es for effect size (correlation)
mode: Mediation model mode: "medsimple" for simple mediation with paths a, b, and cprime, or "medcomplex" for multiple/serial mediator models.
a: The expected standardized effect of the independent variable on the mediator
b: The expected standardized effect of the independent variable on the mediator
cprime: The expected standardized effect of the independent variable on the mediator
model_type: Type of complex mediation model when mode="medcomplex": "twomeds" for two parallel mediators, "threemeds" for three parallel mediators, or "twoserial" for a two-mediator serial model.
a1: The expected standardized effect of the independent variable on mediator 1
b1: The expected standardized effect of the independent variable on mediator 1
a2: The expected standardized effect of the independent variable on mediator 1
b2: The expected standardized effect of the independent variable on mediator 1
a3: The expected standardized effect of the independent variable on mediator 1
b3: The expected standardized effect of the independent variable on mediator 1
d1: The expected standardized effect of the independent variable on mediator 1
d2: The expected standardized effect of the independent variable on mediator 1
r12: The expected standardized effect of the independent variable on mediator 1
r13: The expected standardized effect of the independent variable on mediator 1
r23: The expected standardized effect of the independent variable on mediator 1
cprime2: The expected standardized effect of the independent variable on the mediator
sensitivity_coef: For complex mediation models with the effect-size aim, the path coefficient ("a1", "b1", "d1", ...) that is resized to find the minimum detectable indirect effect.
power: Minimal desired power
n: Sample size
sig.level: Type I error rate (significance cut-off or alpha)
alternative: Test direction: "two.sided" (default) or "one.sided".
test: Mediation test method: "sobel", "joint" for joint-significance, or the Monte Carlo confidence-interval methods "parametric" and "simulation" (both approximate bootstrap-CI power for the indirect effect).
mcR: Number of simulated studies used by the Monte Carlo CI methods.
parallel: Logical; if TRUE, use parallel computation for simulation-based methods when available.
set_seed: Logical; if TRUE, use the value in seed to make simulations reproducible.
seed: Random seed used when set_seed=TRUE.
table_pwbyn: Logical; if TRUE, produce the "Power by Sample size" table.
table_pwbyes: Logical; if TRUE, produce the "Power by Effect Size" table: the indirect effect and the value of the varied coefficient reaching power .5/.8/.95 at the analysis sample size.
plot_ncurve: Logical; if TRUE, produce the power-by-n curve.
plot_log: Logical; if TRUE, use a log scale where supported in plots.
plot_palette: Color palette used for plots.
plot_x: Parameter shown on the X axis of the custom power plot.
plot_y: Parameter shown on the Y axis of the custom power plot.
plot_custom_labels: Logical; if TRUE, add value labels to the custom plot.
plot_z: Parameter used to split the custom power plot into multiple lines or conditions.
plot_x_from: Lower bound of the X-axis parameter range for the custom power plot.
plot_x_to: Upper bound of the X-axis parameter range for the custom power plot.
plot_z_lines: Number of reference lines or levels to use for plot_z in the custom plot.
plot_z_value: Explicit values of the plot_z parameter to use in the custom plot.
plot_to_table: Logical; if TRUE, output the plotted custom power values as a table.
test_c: Logical; if TRUE, also compute the power analysis for the direct X -> Y effect (c path) without mediators.
explain: Logical; if TRUE, include explanatory output where available.
diagram: Logical; if TRUE, display the mediation path diagram.
.interface: Used for internal purposes
.caller: Used for internal purposes

Value

A results object containing:

`results$intro`					a html
`results$extrainfo`					a html
`results$issues`					a html
`results$initnotes`					a html
`results$diagram`					an image
`results$powertab`					a table
`results$effectsize`					a table
`results$powerbyn`					a table
`results$powerxy`					a table
`results$plotnotes`					a html
`results$powerNcurve`					an image
`results$powerCustom`					an image
`results$customnotes`					a html
`results$customtable`					a table

Tables can be converted to data frames with asDF or as.data.frame. For example:

results$powertab$asDF

as.data.frame(results$powertab)