How to Run Mediation Analysis in SPSS

Introduction

Mediation analysis is a statistical method used to explore how and why one variable influences another through an intermediary variable, known as the mediator. It is an essential technique for understanding indirect effects and the underlying mechanisms of relationships between variables. In this guide, we will cover:

• The concept and purpose of mediation analysis.
• The use of traditional regression models for mediation.
• How to implement mediation analysis using the Hayes PROCESS Macro in SPSS.
• A practical example to illustrate mediation analysis.
• Step-by-step instructions for running and interpreting mediation analysis results.

By the end of this guide, you will have a clear understanding of how to conduct and interpret mediation analysis effectively in SPSS.

1. How to Run Mediation Analysis in SPSS

Mediation analysis is a valuable tool for researchers seeking to uncover the mechanisms through which one variable influences another. By incorporating a mediator variable, this method provides insights into the direct and indirect effects within a causal framework. In this post, we explore the fundamental steps and tools required to perform mediation analysis using SPSS. You will learn the basics of mediation analysis, how to use regression models for mediation, and the benefits of leveraging the Hayes PROCESS Macro. Additionally, a practical example and step-by-step guide will help you understand how to conduct and interpret mediation analysis effectively in SPSS.

What is Mediation Analysis?

Mediation analysis examines whether the relationship between an independent variable (X) and a dependent variable (Y) is explained by a third variable (M), the mediator. This method quantifies both direct and indirect effects.

Key Terms:

• Direct effect: The effect of X on Y, excluding the mediator.
• Indirect effect: The effect of X on Y through the mediator (M).
• Total effect: The combined direct and indirect effects.

Purpose of Mediation Analysis: To understand the mechanism or process through which an independent variable influences a dependent variable.

Assumptions:

• Variables should have a causal relationship.
• No multicollinearity among variables.
• Linear relationships between variables.
• Residuals should follow a normal distribution.

Traditional Regression Models for Mediation

Baron and Kenny’s (1986) method involves three regression equations:

1. Path a: Regress the mediator (M) on the independent variable (X).
   M = b0 + b1X + e
2. Path b: Regress the dependent variable (Y) on the mediator (M).
   Y = b0 + b2M + e
3. Path c’ (Direct Effect): Regress the dependent variable (Y) on both the independent variable (X) and the mediator (M).
   Y = b0 + b3X + b2M + e

The indirect effect is calculated as aimesba imes b. If the direct effect (c’) is reduced compared to the total effect (c), partial mediation is present. If c’ is non-significant, full mediation occurs.

Hayes PROCESS Macro

The Hayes PROCESS Macro simplifies mediation analysis by automating calculations and generating detailed output, including bootstrap confidence intervals for indirect effects.

How to Set Up Hayes PROCESS Macro in SPSS:

1. Download and Install: Follow the same installation steps outlined for moderation analysis.
2. Open the Macro: Navigate to “Analyze > Regression > PROCESS vX.X.”
3. Configure the Model:
   • Set the dependent variable (Y), independent variable (X), and mediator (M).
   • Select model 4 (simple mediation).
4. Run the Analysis: Check the output for total, direct, and indirect effects, along with bootstrap confidence intervals.

Advantages of Hayes PROCESS Macro:

• Reduces calculation errors.
• Provides bootstrap confidence intervals for robustness.
• Outputs detailed paths and effect sizes.

2. Example for Mediation Analysis

Research Example: A researcher wants to explore whether job satisfaction mediates the relationship between work environment and employee performance. Here:

• Independent variable (X): Work environment quality.
• Mediator (M): Job satisfaction.
• Dependent variable (Y): Employee performance.

3. Step-by-Step Conducting Mediation Analysis in SPSS

Using Regression Analysis

1. Prepare the Data: Ensure all variables are properly coded and entered.
2. Run Path a: Regress the mediator (M) on the independent variable (X). Save coefficients.
3. Run Path b: Regress the dependent variable (Y) on the mediator (M).
4. Run Path c’: Regress Y on both X and M.
5. Calculate Indirect Effect: Multiply coefficients of Path a and Path b.

Using Hayes PROCESS Macro

1. Open PROCESS: Navigate to “Analyze > Regression > PROCESS v3.5.”
2. Input Variables: Enter Y, X, and M in the respective fields.
3. Configure Options: Select model 4 and enable bootstrap confidence intervals.
4. Run Analysis: Review the output for indirect, direct, and total effects.

4. How to Interpret Results of Analysis

1. Direct and Indirect Effects: Examine coefficients and p-values for each path. A significant indirect effect supports mediation.
2. Bootstrap Confidence Intervals: Check if the confidence interval for the indirect effect excludes zero, confirming mediation.
3. Effect Size: Use the proportion of the indirect effect to the total effect to assess the strength of mediation.
4. Model Summary: Evaluate R² and significance to ensure the model fits well.

By following this guide, you can confidently conduct mediation analysis in SPSS and interpret its results to uncover the mechanisms behind observed relationships.

5. Interpretation of Mediation Analysis Results

The mediation analysis provides insights into how the independent variable (X) affects the dependent variable (Y) through the mediator (M). Below is the detailed interpretation:

Outcome Variable: Mediator (M)

• R-squared (0.1179): About 11.79% of the variance in the mediator (M) is explained by the independent variable (X).
• p-value (0.0005): The model predicting M is statistically significant.
• Coefficient for XX (-0.2849): There is a significant negative relationship between X and M (p=0.0005p = 0.0005). For every unit increase in X, M decreases by 0.2849 on average.

Outcome Variable: Dependent Variable (Y)

• R-squared (0.6322): About 63.22% of the variance in Y is explained by X and M together.
• p-value (< 0.0001): The overall model predicting Y is statistically significant.
• Coefficient for X (-0.1227): The direct effect of X on Y is significant (p<0.0001p < 0.0001). For every unit increase in X, Y decreases by 0.1227 when controlling for M.
• Coefficient for M (-0.3286): The mediator (M) has a significant negative effect on Y (p<0.0001p < 0.0001). For every unit increase in M, Y decreases by 0.3286 on average.

Direct Effect of X on Y

• Effect (-0.1227): This is the direct relationship between X and Y, excluding the influence of M. The direct effect is statistically significant (p<0.0001p < 0.0001).

Indirect Effect of X on Y via M

• Effect (0.0936): The indirect effect measures the influence of X on Y through M. The positive coefficient indicates that the indirect effect contributes to a slight increase in Y.
• Bootstrap Confidence Interval (0.0397, 0.1481): Since the confidence interval does not include zero, the indirect effect is statistically significant. This supports the presence of mediation.

Summary of Mediation

1. Significant Direct Effect: X directly influences Y.
2. Significant Indirect Effect: X affects Y indirectly through M.
3. Partial Mediation: Since both the direct and indirect effects are significant, this is an example of partial mediation, where M explains part of the relationship between X and Y.

The results suggest that while X has a direct effect on Y, a portion of the effect is mediated through MM, offering insights into the mechanisms of this relationship.

Some Useful Pages

• > Dissertation Statistics Tutor
• > Help with Dissertation Statistics
• > APA Statistical Reporting Help
• > Statistics Help for Students
• > Mixed Methods Data Analysis 
• > Quantitative & Qualitative Data Analysis 
• > Statistical Data Analysis for Psychology
• > Expert SPSS Consultant
• > Statistical Consulting for Psychology
• > Statistical Help for Students
• > SPSS Results Interpretation
• > Help with Statistical Methods
• > Likert Scale Data Analysis
• > Experimental Psychology
• > Statistical Analysis Software
• > How to Run Moderation Analysis in SPSS
• > How to Run SEM in SPSS AMOS

Also, connect with us on LinkedIn and YouTube for more updates and valuable content.