Article summary with Baron and Kenny (1986)

This is a summary of the Article of Baron and Kenny (1986)

These are content for the exam of the Research Methods component of Module 7 (Research Methods and Research Project) of the University of Twente, in the Netherlands. 

The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations.

• The article aims to clarify the distinction between moderator and mediator variables on various - levels --> differences both conceptually and strategically
  • Moderator function as dividing an independent variable into subgroups to show its effectiveness regarding a dependent variable
  • Mediator function explains how the independent variable influences the dependent variable

Moderator variable

• Moderators can be qualitative, or quantitative, affecting the strength or direction of this relationship.
• In correlational framework, a moderator is a third variable that alters the correlation between two other variables (e.g. a study found that the correlation between life events and illness severity was stronger for uncontrollable events compared to controllable ones)
• In terms of analysis, a basic moderator effect can be seen as an interaction between an independent variable and a factor specifying its conditions

Analytic procedures for testing moderation

Based on different combinations of continuous and categorical variables

• 1. When both the moderator and independent variables are categorical --> 2x2 ANOVA can reveal moderation through an interaction
• 2. If the moderator is categorical and the independent variable is continuous --> regression coefficients
• 3. When the moderator is continuous and the independent variable is categorical --> hypotheses about the effect change as the moderator varies are tested
• 4. If both moderator and independent variables are continuous --> analysis can include dichotomizing the moderator or considering linear/quadratic moderation

Mediator variable:

• They explain the relationship between an independent variable and a dependent variable
  • Woodworm's S-O-R model: highlights the role of internal processes between stimuli and responses
• Conditions for a variable to function as a mediator:

• • Variations in the independent variable significantly account for variations in the mediator (path c)
  • Variations in the mediator significantly account for variations in the dependent variable (path b)
  • When paths a and b are controlled, the previously significant relation between the independent and dependent variables becomes nonsignificant, with strong mediation indicated when path c is zero
• To test for mediation, three regression equations are used:
  
  • 1. Regress the mediator on the independent variable
  • 2. Regress the dependent variable on the independent variable
  • 3. Regress the dependent variable on both the independent variable and the mediator
  • --> the effect of the independent variable on the dependent variable less in the third equation than in the second, then mediation is established
• Multicollinearity: when the presence of a correlation between the independent variable and the mediator --> reducing the power of the third equation's coefficients
  
  • Dor this, it's essential to examine both the significance and the size of coefficients
    
    • Sobel (1982) approximate significance test for the indirect effect of the independent variable on the dependent variable via the mediator --> use of multiple regression to estimate a mediational model, assuming no measurement error in the mediator and no causal relationship between the dependent variable and the mediator
      
      • However, these assumptions can lead to underestimation of the mediator's effect and overestimation of the independent variable's effect on the dependent variable

Clarifying the Meaning of Control

• Control moderates rather than mediate

Behaviour-intention relation

• Behavioural intention (BI) is seen as a clear mediator

Attitudes and traits

• Self-monitoring as a moderator in predicting behaviour from attitudes or traits
• This framework emphasizes the dynamic role of mediators in influencing behaviour compared to the relatively static classification of moderators.