Class Overview

• Recording: Always record classes and discussions for future reference.

• Communication: Students can email with any questions or concerns before, during, or after class.

Class Approach

• Simplification: The class material will be approached in a straightforward manner, preventing overthinking.

• Class Structure: Slow-paced class with focus on essential material; auxiliary material will be flagged for future quizzes as needed.

Review of Previous Class

• Data Cleaning: Recap on the importance of data cleaning, especially for quantitative analysis, particularly using data obtained from platforms like Qualtrics.

  • Data Cleaning Guidelines: Follow step-by-step instructions provided in class for assignment due in week 8, including:

  • Imputation techniques, specifically EM (Expectation-Maximization).

  • Documentation of cleaning procedures.

  • Ensure data sheets are neat and devoid of errors once cleaned.

Introduction to Regression

• Basic Understanding: Tonight's focus will be on regression, particularly mediation regression.

  • Clarification that regression is a significant analytical method in psychology, particularly useful for understanding relationships between factors in quantitative research.

• Multiple Regression: Acknowledgment of the need for reviewing multiple regression concepts since they have been introduced in undergraduate studies.

  • Online Resources: Students should access the recorded lecture covering basic regression concepts for foundational review.

Mediation Analysis

• Definition: Mediation analysis helps unearth intra-relationships, showing how one variable influences another through a mediator.

  • Differences between mediation and moderation will be clarified in next week's class.

• Path Analysis: The lecture includes path analysis concepts that connect to regression models and structural equation models (SEM).

  • Structural equation modeling is an advanced methodological approach typically stemming from mediation analysis.

  • Understanding how different types of regression (ANOVA, t-tests, correlations) fit into SEM is crucial.

Types of Regression

• Simple Linear Regression: Basic understanding of regression where one predictor influences an outcome.

• Multiple Regression: Understanding how more than one independent variable affects a dependent variable.

  • Discussion of regression techniques will include hierarchical and stepwise regression.

Steps for Tonight's Class Workshop

• Workshop Activities: Actual regression analysis will be performed during class tonight, practically applying these concepts:

  1. Assumption Tests: Importance of fulfilling regression assumptions before analysis.

  2. Utilizing SPSS: Instruction on how to run regression analyses in SPSS, including the utilization of the Hayes PROCESS macro for mediation analysis.

Assumption Testing for Regression

• Essential Assumptions: A series of assumption tests must be carried out before regression analysis, including:

  • Linearity: Checking if relationships between variables are linear.

  • Homoscedasticity: Examining if residuals maintain a constant variance.

  • Multicollinearity: Assessing whether independent variables are correlated with each other.

  • Normality: Ensuring the data approximates a normal distribution.

• Residual Plots: A key tool to visualize these assumptions; it indicates whether assumptions are met.

  • Guidelines on interpreting residual plots will be provided.

Practical Application and Assignment Preparation

• Students should prepare their data sheets and run assumption tests as part of their assignments.

• Reference materials, recaps, and lectures will support them through statistical analysis tasks.

Regression Models and Definitions

• Mediator vs Moderator: A comprehensive definition will be provided,

  • A mediator affects the strength of a relationship between variables.

  • A moderator affects the direction or strength of the relationship without being part of that relationship.

• Paths in Mediation Analysis:

  • A: Predictor variable to mediator.

  • B: Mediator to outcome variable.

  • C: Total effect of the predictor on the outcome variable (C' being direct).

Reporting Results

• Guidelines on effectively reporting mediation results will be provided, including:

  • Indicating significant effects and how to write up findings.

  • Importance of presenting confidence intervals rather than p-values in bootstrapped results.

• Encouragement to critically evaluate and discuss findings within the context of existing literature.

Sample Size Considerations in Mediation Analysis

• Importance of understanding sample size requirements specific to mediation analysis, different from traditional approaches using tools like G-Power.

• Reference works that suggest methodologies for determining appropriate sample sizes tailored to mediation studies (e.g., Fritz and MacKinnon).

Covariate Considerations

• Brief overview of controlling for covariates in regression analysis, including:

  • Why Control?: To isolate the effects of the primary independent variable and avoid confounding results.

  • Adding Covariates: Process of including additional relevant variables in regression analyses.

Final Notes

• The class concludes with a brief discussion on multiple mediation and its relevance. Understanding multiple mediation can enhance the complexity and insight of analysis, granting a broader view of parameter influences.

• Recap on upcoming lectures, specifically the transition into moderation analysis.

Questions and Discussions

• Open floor for further questions, emphasizing the importance of clarifying understanding as it relates to upcoming assignments and broader applications in research.