Mediation and moderation analysis in your browser

Crosstabs fits the regression-based models psychologists usually run with Hayes' PROCESS macro (models 4, 1, 6, 7 and 14) on your device, with seeded percentile bootstrap intervals for indirect effects. It is an independent implementation, not affiliated with Andrew F. Hayes or PROCESS.

• How much of the effect of X on Y runs through each mediator (indirect effects a × b with 95% bootstrap intervals) and how much remains direct (c′).
• Whether two parallel indirect effects differ (pairwise contrasts) and how large they are on a standardized scale (completely standardized indirect effects).
• Whether the effect of X on Y changes with a moderator (interaction test with R² change), the conditional effect at chosen moderator values and the Johnson–Neyman region where it is significant.
• Whether an indirect effect depends on a moderator (index of moderated mediation with a bootstrap interval, plus conditional indirect effects).

1. 01

   Choose the procedure and model

   In Statistics mode open Regression → Mediation and moderation (PROCESS-style), then pick a Model: 4 parallel mediation, 1 simple moderation, 6 serial mediation, 7 or 14 moderated mediation.

2. 02

   Assign the variables

   Choose the Independent variable (X) and Outcome (Y), use Add mediator (M) for each mediator (two, in causal order, for model 6), and pick the Moderator (W) for models 1, 7 and 14. Add covariate enters a control in every equation.

3. 03

   Set the options

   Under Options keep Bootstrap samples at 5000 or change it, note the Random seed, set the Confidence level, and tick HC3 standard errors, the Sobel test or mean-centering if you need them. For moderation choose where to probe W: mean and ±1 SD, the 16th, 50th and 84th percentiles, or custom values.

4. 04

   Declare the weight and run

   If the workspace has a weight, choose Frequency weight or Run unweighted (a sampling weight is blocked with an explanation), then click Run analysis. The bootstrap runs in a background worker.

5. 05

   Report the results

   Copy sentences from the APA-style summary, export any table as CSV, and report the seed and number of resamples so the intervals can be reproduced.

Model Summary and Coefficients (one per outcome)

R, R², MSE and F for every equation, and each path coefficient with its SE, t, p and confidence interval (LLCI, ULCI). With a moderator, the Test of Highest-Order Unconditional Interaction gives the R² change and its F test.

Total, direct and indirect effects of X on Y

The total effect c, the direct effect c′ and each indirect effect with its bootstrap SE and percentile interval (BootLLCI, BootULCI); an interval that excludes zero indicates mediation. Completely standardized effects and pairwise contrasts follow.

Conditional effects and Johnson–Neyman significance region

The effect of the focal predictor at each probed moderator value, the moderator values where significance changes (with the share of cases below and above), a table across the moderator's range and a simple-slopes plot.

Index of moderated mediation

For models 7 and 14: the index (a₃ × b or a × b₃) with its bootstrap interval, and the conditional indirect effect at each moderator value. An index interval excluding zero means the indirect effect depends on W.