Get AI summaries of any video or article — Sign up free
Mediation Analysis with #ChatGPT and Hayes Process Macro using #SPSS  (Model 4) thumbnail

Mediation Analysis with #ChatGPT and Hayes Process Macro using #SPSS (Model 4)

Research With Fawad·
5 min read

Based on Research With Fawad's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

Culture significantly predicts commitment, supporting the first step of the mediation chain.

Briefing

Mediation analysis with SPSS PROCESS Model 4 is used to test whether “culture” affects “organizational performance” indirectly through “commitment,” rather than acting only through a direct link. The core question is whether the effect of the independent variable (culture) on the dependent variable (performance) passes through a mediator (commitment). In this setup, culture is expected to influence commitment, and commitment is expected to influence performance; if both pathways are statistically supported, the indirect effect can be quantified as the product of the “a” path (culture → commitment) and the “b” path (commitment → performance).

The workflow starts in SPSS by running PROCESS Model 4 with culture as the independent variable, commitment as the mediator, and organizational performance as the dependent variable. The analysis requests standardized effects and total effects, then inspects the output in two regression stages. First, culture predicts commitment in a regression where culture is the only predictor. The reported correlation between culture and commitment is R = 0.609, yielding R² ≈ 0.378 (about 37.8% of variance in commitment explained by culture). The culture → commitment path is statistically significant (p < .05), with a t statistic exceeding 1.96 and a “good” coefficient.

Second, organizational performance is regressed on both culture and commitment, producing R² ≈ 0.431 (about 43.1% of variance in performance explained by the two predictors). Both culture and commitment are significant predictors of performance (again, p < .05 and no zero in the reported confidence interval bounds), indicating that culture has an effect on performance even when commitment is included, and commitment also contributes uniquely.

The analysis then distinguishes among three effects. The total effect combines direct and indirect components. The direct effect is the culture → performance relationship while commitment remains in the model (denoted as C′). The indirect effect is computed as a × b, where a is the culture → commitment coefficient and b is the commitment → performance coefficient. Using the reported coefficients, the indirect effect is calculated as 0.641 × 0.4530 ≈ 0.2736, and it is significant because the confidence interval for the indirect effect does not cross zero.

Because both the direct effect (culture → performance with commitment controlled) and the indirect effect (culture → commitment → performance) are significant, the mediation pattern is classified as partial mediation. In practical terms, culture influences organizational performance through commitment, but it also retains a separate direct influence on performance.

A key methodological takeaway is how interpretation is handled: results from PROCESS output should be converted into a clean, structured form before using AI for write-ups. The session emphasizes that AI can help draft APA-style reporting, but meaningful conclusions still depend on understanding which coefficients correspond to the a, b, direct, indirect, and total effects—then interpreting significance using confidence intervals and p-values rather than copying raw output blindly.

Cornell Notes

The mediation test uses SPSS PROCESS Model 4 to examine whether “culture” affects “organizational performance” through “commitment.” Culture significantly predicts commitment (R = 0.609; R² ≈ 0.378), and both culture and commitment significantly predict performance (R² ≈ 0.431). The indirect effect of culture on performance through commitment is computed as a × b (0.641 × 0.4530 ≈ 0.2736) and is significant because its confidence interval does not include zero. The direct effect of culture on performance remains significant even with commitment in the model, so the pattern is partial mediation. This matters because it clarifies that culture shapes performance both directly and indirectly via commitment.

What does PROCESS Model 4 test in a mediation setup?

Model 4 tests whether the independent variable (culture) influences the dependent variable (organizational performance) indirectly through a mediator (commitment). It estimates (1) the a path: culture → commitment, (2) the b path: commitment → performance, and (3) the direct effect (C′): culture → performance while commitment is included. The indirect effect is quantified as a × b, and the total effect combines direct and indirect components.

How are direct, indirect, and total effects distinguished in this analysis?

The direct effect (C′) is the culture → performance relationship when commitment is present in the regression. The indirect effect is the product of the two paths that form the mediation chain: a (culture → commitment) multiplied by b (commitment → performance). The total effect is the combined influence of direct and indirect effects, expressed as C + (a × b) (or equivalently as the sum of the direct and indirect components reported by PROCESS).

What evidence supports mediation here?

Mediation requires that the indirect effect is significant and that the relevant paths are significant. Culture significantly predicts commitment (p < .05; t > 1.96), and commitment significantly predicts performance (p < .05; confidence interval bounds exclude zero). The indirect effect is calculated as 0.641 × 0.4530 ≈ 0.2736 and is significant because the confidence interval for the indirect effect does not cross zero.

Why is the mediation labeled “partial” rather than “full”?

Partial mediation occurs when both the indirect effect is significant and the direct effect remains significant. In this analysis, the direct effect of culture on performance (with commitment included) is significant, and the indirect effect through commitment is also significant. If the direct effect were not significant while the indirect effect was, that would indicate full mediation.

How should significance be interpreted using the output described here?

Significance is assessed using p-values (p < .05) and confidence intervals. The session repeatedly notes that confidence intervals for the relevant coefficients do not include zero, which supports statistical significance. For the indirect effect specifically, the confidence interval for a × b must exclude zero to claim a significant mediation effect.

Review Questions

  1. What are the a and b paths in this mediation model, and how is the indirect effect computed?
  2. If the indirect effect is significant but the direct effect is not, what mediation type would that indicate?
  3. Which reported statistics (R, R², p-values, confidence intervals) are used to justify that culture and commitment significantly predict organizational performance?

Key Points

  1. 1

    Culture significantly predicts commitment, supporting the first step of the mediation chain.

  2. 2

    Commitment significantly predicts organizational performance when culture is included, supporting the second step of the chain.

  3. 3

    The indirect effect of culture on performance through commitment is computed as a × b and is significant (confidence interval excludes zero).

  4. 4

    The direct effect of culture on performance remains significant even after including commitment, indicating partial mediation.

  5. 5

    R² values show that culture alone explains about 37.8% of variance in commitment, while culture plus commitment explains about 43.1% of variance in performance.

  6. 6

    Interpreting mediation requires mapping coefficients to a, b, direct, indirect, and total effects rather than relying on raw PROCESS output alone.

  7. 7

    AI can assist with APA-style write-ups, but only after the relevant mediation quantities are correctly identified and formatted.

Highlights

The indirect effect is calculated as a × b: 0.641 × 0.4530 ≈ 0.2736, and it is significant because its confidence interval does not include zero.
Both direct and indirect effects are significant, so the model shows partial mediation rather than full mediation.
Culture explains about 37.8% of the variance in commitment (R² ≈ 0.378), while culture plus commitment explains about 43.1% of variance in performance (R² ≈ 0.431).
The analysis distinguishes direct effect (C′) from indirect effect (a × b) to determine whether commitment carries part of culture’s influence on performance.

Mentioned