The best PROCESS macro alternative for running mediation and moderation online is a browser-based tool that performs bootstrapped indirect effects and interaction analyses without SPSS or Hayes' installed macro — StatRyx, JASP, jamovi, and R's lavaan and mediation packages all do this, with StatRyx being the easiest for non-statisticians who also need the APA write-up done automatically. If you've ever fought with a broken PROCESS .sps file, a licensing wall, or an error telling you your model number doesn't exist, you already know why people go looking for something simpler.
Key Takeaways
- PROCESS is a free macro by Andrew F. Hayes that runs inside SPSS, SAS, or R — it is not standalone software, so you still need a (paid) SPSS license or coding skills to use it.
- The main online alternatives are StatRyx, JASP, jamovi, and R packages (
lavaan,mediation) — all can estimate mediation, moderation, and moderated mediation with bootstrapped confidence intervals. - A significant indirect effect in mediation is confirmed when the bootstrapped 95% CI does not contain zero — not by the Sobel test, which is now considered outdated.
- StatRyx runs mediation and moderation entirely in the browser and returns APA 7-formatted output, so you skip both the macro install and the manual write-up.
- SPSS licenses cost roughly $99+ per month for individuals, which is the single biggest reason researchers seek a PROCESS alternative.
What is the PROCESS macro, and why look for an alternative?
The PROCESS macro is a free add-on written by Andrew F. Hayes that estimates mediation, moderation, and conditional process (moderated mediation) models inside SPSS, SAS, or R using a numbered template system. It became the default tool in psychology and social science because it automates bootstrapping for indirect effects — something base SPSS does not do out of the box.
The catch is that PROCESS is a macro, not a program. To use it you need host software, and in most labs that means SPSS, which costs around $99 per month or roughly $1,200+ per year for an individual subscription. On top of the price, three recurring pain points send people searching for an alternative:
- Installation headaches — the .sps or .R file has to be loaded every session, and version mismatches break it.
- The model-number maze — you have to memorise which of Hayes' ~90 numbered templates matches your diagram.
- No automatic write-up — PROCESS gives you raw output tables, and you still translate every coefficient into APA prose yourself.
An online alternative solves all three: nothing to install, no model numbers to look up, and (with the right tool) a finished results paragraph.
What are the best online PROCESS macro alternatives?
Here are the realistic options, ranked for a researcher who is not a programmer but needs correct, publishable results.
1. StatRyx — best for non-statisticians who want the write-up done
StatRyx is an AI-powered statistical analysis tool that runs mediation and moderation entirely in your browser and returns APA 7-formatted output. You upload your data, pick your predictor, mediator/moderator, and outcome, and it runs bootstrapped indirect effects and interaction terms, then writes the results paragraph for you. Pros: no install, no model numbers, automatic APA reporting, chooses the correct model structure. Cons: newer than the incumbents; cloud-based, so very large proprietary datasets may need an institutional review.
2. JASP — best free desktop option
JASP offers mediation and moderation through its SEM and regression modules with a clean point-and-click interface. Pros: free, open-source, Bayesian options. Cons: it's a desktop install, not online, and moderated mediation still leans on the SEM module, which has a learning curve.
3. jamovi — best free option with a PROCESS-style add-on
jamovi has the medmod and jAMM modules that replicate much of PROCESS directly. Pros: free, familiar spreadsheet layout, dedicated mediation module. Cons: desktop install; conditional process models require the add-on and some setup.
4. R (lavaan and mediation packages) — most powerful, most technical
R gives you complete control over indirect effects and bootstrapping. Pros: free, publication-grade, infinitely flexible, fully reproducible. Cons: code-heavy — you'll write model syntax and interpret raw output yourself, which is a real barrier if you're not a programmer.
5. SPSS + PROCESS — the incumbent
The original setup. Pros: what most supervisors know; Hayes' documentation is excellent. Cons: SPSS is paid (~$99+/month), PROCESS must be reinstalled per session, and there's no automatic APA write-up.
PROCESS macro alternatives compared
| Tool | Online / Install | Cost | Needs coding? | Auto APA write-up |
|---|---|---|---|---|
| StatRyx | Online (browser) | Free tier | No | Yes |
| JASP | Desktop install | Free | No | No |
| jamovi | Desktop install | Free | No | No |
R (lavaan/mediation) |
Install + code | Free | Yes | No |
| SPSS + PROCESS | Desktop install | ~$99+/mo | No | No |
How do I run a mediation analysis without PROCESS?
To run a mediation analysis without PROCESS, you estimate whether an independent variable (X) affects an outcome (Y) through a mediator (M), then test the indirect effect (a × b) with bootstrapped confidence intervals. Any of the tools above does this; the logic is identical to PROCESS Model 4.
Worked example: does study time affect exam scores through confidence?
Imagine 120 students. You hypothesise that study hours (X) raise exam score (Y) partly because they increase confidence (M).
- Path a (X → M): more study hours predict higher confidence, b = 0.42, p < .001.
- Path b (M → Y): confidence predicts exam score controlling for study hours, b = 0.55, p = .002.
- Indirect effect (a × b): 0.42 × 0.55 = 0.23.
- Bootstrap it (5,000 resamples): 95% CI [0.09, 0.38].
Because the confidence interval does not include zero, the indirect effect is significant — confidence partially mediates the study-time effect. You'd report it like this:
The indirect effect of study hours on exam score through confidence was significant, b = 0.23, 95% bootstrap CI [0.09, 0.38].
Note what we did not use: the Sobel test. Modern practice relies on bootstrapped CIs because indirect effects aren't normally distributed, and the CI method has more power.
How do I run a moderation analysis online?
Moderation tests whether the strength of the X → Y relationship changes depending on a third variable (W), and online you test it by adding an X × W interaction term to a regression and checking whether that term is significant. This mirrors PROCESS Model 1.
For example, if the effect of study hours on exam score is stronger for students with high motivation, your interaction term might come out as b = 0.18, t(116) = 2.41, p = .017. A significant interaction means motivation moderates the relationship — and you'd probe it with simple slopes at low, mean, and high motivation. StatRyx generates the simple-slopes table and