PSY4101 Assignment 2: Hierarchical Regression & Moderation in Jamovi Tutorial

Introduction: Why This Matters for Your PSY4101 Assignment

In the 2025-2026 academic year, data analysis skills are more critical than ever. Whether you're analyzing student performance, consumer behavior, or even AI model predictions, understanding how to control for variables and test complex hypotheses is essential. This tutorial walks you through the key concepts needed for PSY4101 Assignment 2, focusing on hierarchical regression, moderation, and mediation using jamovi. We'll use the assignment's scenario—exam performance predicted by personality, anxiety, and revision—to illustrate each step.

Checking Regression Assumptions: Linear Independence and Normality

Before running your model, you must verify assumptions. In Problem 1A(i), you're asked about the linear-independence assumption (no perfect multicollinearity) and the normality assumption (residuals are normally distributed).

Linear Independence (No Perfect Multicollinearity)

This assumption means that no predictor is a perfect linear combination of others. In practice, check the Variance Inflation Factor (VIF). In jamovi, after running the regression, look at the 'Collinearity Statistics' table. VIF values above 10 (or sometimes 5) indicate problematic multicollinearity. For the Big Five personality scores, they are often moderately correlated but not perfectly. If VIF is below 5 for all predictors, the assumption is satisfied. Report the highest VIF value.

Normality of Residuals

Examine the Q-Q plot of residuals and the Shapiro-Wilk test. In jamovi, under 'Assumption Checks', select 'Normality Test'. If the p-value is > 0.05, the residuals are normal. However, with N=103, minor deviations are acceptable due to the Central Limit Theorem. If violated, you can still proceed but note the limitation.

Interpreting the Regression Equation and Coefficients

After running the regression with exam scores as outcome and the five personality traits as predictors, jamovi provides coefficients. For example:

Exam Score = 50.234 + 2.145*(Openness) - 1.032*(Conscientiousness) + ...

Round to three decimal places. The smallest p-value indicates the strongest predictor. Suppose Conscientiousness has p = 0.002. You would write: "Conscientiousness significantly predicted exam scores (β = 2.145, p = .002), indicating that a one-unit increase in conscientiousness is associated with a 2.145-point increase in exam score, holding other personality traits constant."

Testing Overall Model Significance and R-squared

The ANOVA table tests whether the model explains significant variance. If F(5,97) = 4.567, p < .001, the model is significant. Report R² = 0.190, meaning personality explains 19.0% of the variance in exam scores.

Adding Anxiety: Hierarchical Regression

Now, add anxiety to the model. In jamovi, you can run a hierarchical regression by entering personality in Block 1 and anxiety in Block 2. Look at the R² change. If ΔR² = 0.035, F(1,96) = 4.210, p = .043, then anxiety provides additional explanatory power over and above personality. Report: "Anxiety significantly improved the model, accounting for an additional 3.5% of the variance in exam scores (ΔR² = .035, p = .043)."

Semipartial Correlation: What to Expect

Given that anxiety adds unique variance, the semipartial (part) correlation between anxiety and exam scores (controlling for personality) will be non-zero. In jamovi, you can request 'Part Correlations' in the regression output. Expect a value around 0.187 (sqrt(0.035)). Verify and report.

Moderation Analysis: Revision Duration as a Moderator

Problem 1B(i) asks whether revision duration moderates the anxiety-exam relationship. In jamovi, use the 'Moderation' module. Create an interaction term (Anxiety × Revision). If the interaction is significant (p < .05), moderation exists. For simple slopes, examine the effect of anxiety at low (-1 SD), mean, and high (+1 SD) revision. For example:

At low revision: anxiety negatively predicts exam scores (β = -3.210, p = .001).
At high revision: anxiety is not significant (β = -0.450, p = .450).

This supports the hypothesis that revision weakens the negative effect of anxiety.

Mediation Analysis: Anxiety as Mediator

Problem 1B(ii) tests whether anxiety mediates the effect of neuroticism on revision duration. In jamovi, use the 'Mediation' module. Report the indirect effect (a*b) and its confidence interval. If the direct effect (c') is not significant, it's full mediation; if still significant, partial mediation. Check the sign: if a*b and c' have the same sign, it's consistent mediation; if opposite, inconsistent.

Final Tips for Your Assignment

Remember to include jamovi syntax in your submission. Enable syntax mode via File → Preferences → Syntax. Copy the generated R code. Also, include a use of generative AI declaration. Even if you didn't use AI, state so. This is worth 50% of your mark if omitted.

By following this guide, you'll be able to tackle the hierarchical regression, moderation, and mediation components of PSY4101 Assignment 2 with confidence. Good luck!