Free tool that turns the paths of a simple mediation model (a, b, c′ and c, as they come out of PROCESS by Hayes, lavaan or jamovi) into an APA 7 or Vancouver table, a downloadable PNG diagram of the model and the results paragraph in English and Spanish. The bootstrap CI of the indirect effect is pasted from the original output (resampling cannot be reproduced from the coefficients); without it, the tool computes the Sobel test (z = a·b / sqrt(b²·SE_a² + a²·SE_b²); Sobel, 1982) as a clearly labelled, inferior approximation.
Worked example: social support → self-esteem → life satisfaction, N = 210. With a = 0.42 (SE = 0.11, p < .001) and b = 0.51 (SE = 0.13, p < .001), the indirect effect is a×b = 0.21 with a 95% bootstrap CI of [0.08, 0.29]: there is mediation, because the interval does not include zero. The direct effect is c′ = 0.18 (p = .134), the total effect c = 0.39 (p = .001), and the proportion mediated 0.21/0.39 = .55 (about 55% of the total effect). Without the bootstrap, the Sobel test would give z = 2.74, p = .006.
Why is the bootstrap CI better than the Sobel test? Because Sobel assumes a×b is normally distributed and the product of two coefficients is skewed; the bootstrap respects that skew and has more power (MacKinnon et al., 2002). Can you claim full mediation when c′ is no longer significant? Better not: those labels depend on sample size, and Hayes (2022) recommends reporting the indirect effect with its CI plus the direct and total effects, without labels.
Turn your simple mediation (PROCESS, lavaan, jamovi) into an APA 7 or Vancouver table, PNG path diagram and results paragraph, with bootstrap CI or Sobel test.