p-value from three-variable mediation analysis simulation

Simple 3-variable mediation analysis simulation to test the hypothesis that X -> Y is mediated by the relationship X -> M -> Y. Currently, M and Y are assumed to be continuous variables with Gaussian errors, while X may be continuous or dichotomous.

Usage

p_mediation(
  n,
  a,
  b,
  cprime,
  dichotomous.X = FALSE,
  two.tailed = TRUE,
  method = "wald",
  sd.X = 1,
  sd.Y = 1,
  sd.M = 1,
  gen_fun = gen_mediation,
  return_analysis = FALSE,
  ...
)

gen_mediation(
  n,
  a,
  b,
  cprime,
  dichotomous.X = FALSE,
  sd.X = 1,
  sd.Y = 1,
  sd.M = 1,
  ...
)

Arguments

n

total sample size unless dichotomous.X = TRUE, in which the value represents the size per group

a

regression coefficient for the path X -> M

b

regression coefficient for the path M -> Y

cprime

partial regression coefficient for the path X -> Y

dichotomous.X

logical; should the X variable be generated as though it were dichotomous? If TRUE then n represents the sample size per group

two.tailed

logical; should a two-tailed or one-tailed test be used?

method

type of inferential method to use. Default uses the Wald (a.k.a., Sobel) test

sd.X

standard deviation for X

sd.Y

standard deviation for Y

sd.M

standard deviation for M

gen_fun

function used to generate the required two-sample data. Object returned must be a data.frame with the columns "DV" and "group". Default uses gen_mediation to generate conditionally Gaussian distributed samples. User defined version of this function must include the argument ...

return_analysis

logical; return the analysis object for further extraction and customization?

...

additional arguments to be passed to gen_fun. Not used unless a customized gen_fun is defined

Value

a single p-value

See also

gen_mediation

Author

Phil Chalmers rphilip.chalmers@gmail.com

Examples

# joint test H0: a*b = 0
p_mediation(50, a=sqrt(.35), b=sqrt(.35), cprime=.39)
#> [1] 9.263932e-05
p_mediation(50, a=sqrt(.35), b=sqrt(.35), cprime=.39, dichotomous.X=TRUE)
#> [1] 6.110203e-09

# return analysis model
p_mediation(50, a=sqrt(.35), b=sqrt(.35), cprime=.39, return_analysis=TRUE)
#> lavaan 0.6-21 ended normally after 1 iteration
#> 
#>   Estimator                                         ML
#>   Optimization method                           NLMINB
#>   Number of model parameters                         5
#> 
#>   Number of observations                            50
#> 
#> Model Test User Model:
#>                                                       
#>   Test statistic                                 0.000
#>   Degrees of freedom                                 0

# \donttest{

  # power to detect mediation
  p_mediation(n=50, a=sqrt(.35), b=sqrt(.35), cprime=.39) |>
    Spower(parallel=TRUE, replications=1000)
#> 
#> Execution time (H:M:S): 00:00:20
#> Design conditions: 
#> 
#> # A tibble: 1 × 4
#>       n cprime sig.level power
#>   <dbl>  <dbl>     <dbl> <lgl>
#> 1    50   0.39      0.05 NA   
#> 
#> Estimate of power: 0.996
#> 95% Confidence Interval: [0.992, 1.000]

  # sample size estimate for .95 power
  p_mediation(n=interval(50,200), a=sqrt(.35), b=sqrt(.35), cprime=.39) |>
    Spower(power=.95, parallel=TRUE)
#> 
#> Execution time (H:M:S): 00:23:37
#> Design conditions: 
#> 
#> # A tibble: 1 × 4
#>       n cprime sig.level power
#>   <dbl>  <dbl>     <dbl> <dbl>
#> 1    NA   0.39      0.05  0.95
#> 
#> Estimate of n: 66.0
#> 95% Predicted Confidence Interval: [NA, NA]

# }