This page documents the release history of PROCESS as it has evolved over time. It describes new features added since original release, modifications in operation relative to prior releases, known bugs, and some bug fixes. When new features are added, they are documented in the addendum to the documentation that comes with the PROCESS zip archive. Always use the most current release of PROCESS.
Version 4.3.1 (released 9 April 2023)
This is a minor update that restores the printback option in SPSS to its on position after PROCESS is executed. In the R version, this update prevents PROCESS from executing when one of the variables in the command is not in numeric form. In the SAS version, this update fixes a bug that prevented PROCESS from printing a test of X by M interaction in a mediation model when Y is dichotomous.
This is a minor update that restores the printback option in SPSS to its on position after PROCESS is executed. In the R version, this update prevents PROCESS from executing when one of the variables in the command is not in numeric form. In the SAS version, this update fixes a bug that prevented PROCESS from printing a test of X by M interaction in a mediation model when Y is dichotomous.
Version 4.3 (released 11 February 2023)
A new exclude option (availably only through syntax) allows the user to selectively exclude cases from the analysis identified by row number in the data.
A new exclude option (availably only through syntax) allows the user to selectively exclude cases from the analysis identified by row number in the data.
Version 4.2 (released 13 November 2022)
Counterfactually-defined direct, indirect, and total effects in a mediation model assuming X ("treatment") by M (mediator) interaction can now be generated in model 4 with the xmint option. This option replaces model 74 that was eliminated temporarily in version 3 and then permanently in version 4.
Counterfactually-defined direct, indirect, and total effects in a mediation model assuming X ("treatment") by M (mediator) interaction can now be generated in model 4 with the xmint option. This option replaces model 74 that was eliminated temporarily in version 3 and then permanently in version 4.
Version 4.1 (released 20 April 2022)
A new model 0 was added (available only through syntax) for regression analysis with no moderation or mediation component. New output is now available indicating which cases were deleted prior to analysis. A new option for inference about a weighted sum of regression coefficients is not available with the linsum option in syntax. A bootstrap confidence interval for the sum of direct and indirect effects in a mediation model can now be generated with the total option for some models in which it is not possible to estimate the total effect in the usual way by regressing Y on X without the mediators.
A new model 0 was added (available only through syntax) for regression analysis with no moderation or mediation component. New output is now available indicating which cases were deleted prior to analysis. A new option for inference about a weighted sum of regression coefficients is not available with the linsum option in syntax. A bootstrap confidence interval for the sum of direct and indirect effects in a mediation model can now be generated with the total option for some models in which it is not possible to estimate the total effect in the usual way by regressing Y on X without the mediators.
Version 4.0 (released 21 August 2021)
A few changes to the code have resulted in improvements in performance in all versions. This version of PROCESS dramatically speeds up the estimation of logistic regression models when Y is dichotomous. Partially standardized effects of X are no longer generated when X is continuous, and Model 74 has been permanently eliminated. The effsize option has been replaced by the stand option. If you are interested in standardized effects in mediation-only models, use the stand option now. This is also the first official release of PROCESS for R, documented in the third edition of Introduction to Mediation, Moderation, and Conditional Process Analysis.
A few changes to the code have resulted in improvements in performance in all versions. This version of PROCESS dramatically speeds up the estimation of logistic regression models when Y is dichotomous. Partially standardized effects of X are no longer generated when X is continuous, and Model 74 has been permanently eliminated. The effsize option has been replaced by the stand option. If you are interested in standardized effects in mediation-only models, use the stand option now. This is also the first official release of PROCESS for R, documented in the third edition of Introduction to Mediation, Moderation, and Conditional Process Analysis.
PROCESS version 2 was discontinued with the release of version 3 and is no longer distributed. Version 2 history and known bugs and fixes can be found in a document here. PROCESS version 3 was discontinued with the release of version 4 and is no longer distributed. Version 3 history and known bugs and fixes can be found in a document here.