Mediation in SPSS

How mediation is actually done in SPSS

Base SPSS has no dedicated mediation command, so most people use the PROCESS macro (by Andrew F. Hayes). The typical steps:

1. Download PROCESS and install it as a custom dialog or load the syntax file.
2. Choose the right model number (Model 4 for simple mediation) and assign your X, M and Y variables.
3. Turn on bootstrapping (commonly 5,000 samples) to get a confidence interval for the indirect effect — the recommended test, rather than the older Baron-Kenny causal-steps logic or the Sobel test.
4. Read the indirect effect and its bootstrap confidence interval: if the interval excludes zero, the indirect effect is significant.

Why it is fiddly

You have to install and maintain an external macro, match your hypothesis to the correct model number, and remember to switch on bootstrapping with enough samples and the interval type you want. It works well once set up, but the setup is the part that trips people up — and the output is separate from where the data was collected and cleaned.

A bigger caution is interpretation: a significant indirect effect from cross-sectional data does not establish a causal chain. Estimating mediation from a single time point can produce badly biased estimates of the longitudinal process you actually care about (Maxwell & Cole, 2007). Report it as an association and be careful with causal language unless your design supports it.

How Explonia does it

Explonia runs mediation with bootstrapped indirect effects and confidence intervals built in — no macro to install and no model-number lookup. You select X, M and Y, run it, and read the indirect effect with its bootstrap CI. The survey, data and analysis are in one place, and the “Article description” drafts a first APA paragraph from your own estimates. See the APA reporting guide for mediation. Bootstrap intervals are stochastic, so the exact bounds vary slightly between runs.

Frequently asked questions