How to Run a serial/sequential mediation in SPSS
Based on Research and Analysis's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.
Sequential mediation tests whether an IV affects a DV through a two-step mediator chain rather than a single intermediate variable.
Briefing
Sequential mediation in SPSS lets researchers test a chain of mechanisms—an independent variable (IV) affects a first mediator, which then affects a second mediator, which finally predicts the dependent variable (DV). In the example used here, CSR (labeled “csr emp”) is linked to OCBO through two mediators in sequence: perceived organizational support (POS) and effective commitment. The practical payoff is clear: the analysis doesn’t just ask whether CSR predicts OCBO, but whether CSR’s influence runs through specific psychological pathways, including a step-by-step indirect route.
The tutorial frames sequential mediation as a model with one IV and one DV, but multiple variables in between. Whether a variable belongs in the mediation chain comes down to theory and prior literature. The approach then tests three mediation hypotheses: (1) POS alone mediates the CSR → OCBO relationship; (2) effective commitment alone mediates CSR → OCBO; and (3) POS and effective commitment operate sequentially, with POS first and effective commitment second, mediating CSR → OCBO.
In SPSS, the workflow relies on the PROCESS macro (a prerequisite for running these mediation models). After loading the dataset with precomputed scale averages for CSR, POS, effective commitment, and OCBO, the analysis is set up in PROCESS under regression. The key modeling choice is selecting the sequential mediation model number (model 6), rather than the simpler single-mediator model (noted as model 4). The output is then checked in two layers: the “model effects” (the regression paths among IV, mediators, and DV) and the “total/direct/indirect effects” section that corresponds to mediation hypotheses.
Path checks show that CSR has significant positive effects on POS (the first mediator) and on effective commitment (the second mediator). POS also significantly predicts effective commitment, confirming the required ordering for sequential mediation. The direct path from POS to OCBO is not significant: the confidence interval includes zero, indicating POS alone does not reliably predict OCBO when accounting for the full chain. By contrast, effective commitment significantly predicts OCBO, with a confidence interval that excludes zero.
The mediation conclusions come from the indirect effects. The POS-only mediation hypothesis fails because the confidence interval crosses zero (one bound negative, the other positive), so POS cannot be confirmed as a mediator between CSR and OCBO on its own. The effective commitment-only mediation hypothesis succeeds: its indirect effect is significant with a confidence interval that does not include zero. Most importantly, the sequential mediation hypothesis—POS followed by effective commitment—also holds. The sequential indirect effect is reported as 0.0423, and its confidence interval excludes zero, supporting the claim that CSR influences OCBO through a two-step psychological pathway (POS → effective commitment → OCBO).
Cornell Notes
Sequential mediation tests whether an IV affects a DV through two mediators in a specific order. Using SPSS with the PROCESS macro, the model is set to the sequential mediation option (model 6) and then evaluated via path coefficients and indirect effects. In the example, CSR significantly predicts perceived organizational support (POS) and effective commitment, and POS significantly predicts effective commitment. POS alone does not mediate CSR’s effect on OCBO because its indirect effect’s confidence interval includes zero. Effective commitment does mediate, and POS plus effective commitment sequentially mediate the CSR → OCBO relationship, with the sequential indirect effect reported as 0.0423 and a confidence interval that excludes zero.
What distinguishes sequential mediation from single-mediator mediation in SPSS/PROCESS terms?
How do the “model effects” paths determine whether the mediation chain is plausible?
Why does POS fail as a mediator even though it predicts effective commitment?
What evidence supports effective commitment as a mediator?
What specific result confirms sequential mediation (POS → effective commitment)?
Review Questions
- In a sequential mediation model, which three regression relationships must be significant to justify the mediator ordering (mediator 1 → mediator 2, and IV → each mediator)?
- How does a confidence interval that includes zero change the interpretation of a mediation path or indirect effect?
- Why can POS predict effective commitment but still fail to mediate the IV → DV relationship on its own?
Key Points
- 1
Sequential mediation tests whether an IV affects a DV through a two-step mediator chain rather than a single intermediate variable.
- 2
Theory and literature determine which variables belong in the mediation sequence; the statistical model alone can’t justify the ordering.
- 3
Running sequential mediation in SPSS requires the PROCESS macro, with model 6 selected for the two-mediator sequential setup.
- 4
Significant path coefficients among CSR → POS, CSR → effective commitment, and POS → effective commitment support the feasibility of the sequential mechanism.
- 5
POS alone did not mediate CSR’s effect on OCBO because the POS-only indirect effect confidence interval included zero.
- 6
Effective commitment mediated CSR’s effect on OCBO, supported by a significant indirect effect with a confidence interval excluding zero.
- 7
The sequential indirect effect (POS → effective commitment) was significant (0.0423) with a confidence interval that did not include zero, confirming sequential mediation.