QRM Exam Survival Masterclass
QRM EXAM
SURVIVAL MASTERCLASS
The Complete Exam-Solving Handbook
Prof. Goedele Dierckx
Quantitative Research Methods 2025–2026
Exam Duration: 2.5 hours │ Tools: SPSS + Excel
MLR │ Time Series
Based on reverse-engineering of all available past exams and official solutions
Rania El Ghalbzouri Page 1
MBA Bridging Programme
, QRM Exam Survival Masterclass
STEP 1: EXAM ANALYSIS
Analysis of all provided exams reveals clear patterns. The following frequency table ranks every recurring
question type by how often it appeared and its probability of appearing on your exam.
1.1 Question Type Frequency Table
Question Type Freq Importance Exam Prob.
Coefficient interpretation (incl. dummies, logs) 12+ ESSENTIAL 100%
Write theoretical model equation + assumptions 8+ ESSENTIAL 100%
Joint F-test (R² change) 7+ ESSENTIAL 95%
Omitted variable bias analysis 6+ ESSENTIAL 90%
Residual plot interpretation 6+ ESSENTIAL 90%
Dummy variable creation + interpretation 8+ ESSENTIAL 100%
Interaction effect interpretation 6+ ESSENTIAL 85%
VIF / Tolerance / QMC check 5+ HIGH YIELD 80%
Auxiliary regression model 5+ HIGH YIELD 80%
White test (write auxiliary model) 3+ HIGH YIELD 70%
Log-level / log-log interpretation 6+ ESSENTIAL 90%
Dickey–Fuller test for stationarity 6+ ESSENTIAL 95%
AR(1) model + multiplier effects 4+ HIGH YIELD 80%
Distributed lag model + lag elimination 4+ HIGH YIELD 80%
Autocorrelation test (LM / Breusch–Godfrey) 5+ ESSENTIAL 85%
Short-run vs long-run effects 4+ HIGH YIELD 80%
R² / Adjusted R² interpretation 5+ HIGH YIELD 75%
Residual calculation for specific observation 3+ NICE TO KNOW 50%
Spurious regression concept 2+ NICE TO KNOW 40%
Stationarity of first differences 4+ HIGH YIELD 75%
Rania El Ghalbzouri Page 2
MBA Bridging Programme
, QRM Exam Survival Masterclass
1.2 Recurring Formula Summary
Formula Used For Priority
F = (R²full − R²restricted) / J ÷ (1 − R²full) / (n − k − 1) Joint F-test ESSENTIAL
F = (SSER − SSE) / J ÷ SSE / (n − k − 1) Joint F-test (alt) ESSENTIAL
Total multiplier = β / (1 − γ) AR(1) long-run effect ESSENTIAL
Tolerance = 1 − R²aux Multicollinearity check HIGH YIELD
VIF = 1 / Tolerance Multicollinearity check HIGH YIELD
Test stat = nR²aux (χ², df = 1) LM autocorrelation test ESSENTIAL
Test stat = nR²aux (χ², df = p) White test HIGH YIELD
Log interpretation: β/100 for 1% change Log-level model ESSENTIAL
Dummy log: approx (eβ − 1) × 100% Log-dummy effect HIGH YIELD
Critical value DF with trend: t* = −3.41 Dickey–Fuller (5%) ESSENTIAL
Critical value DF no trend: t* = −2.86 Dickey–Fuller (5%) ESSENTIAL
1.3 Recurring SPSS Procedures
SPSS Task Menu Path Priority
Linear Regression Analyze → Regression → Linear ESSENTIAL
R² Change (Joint F-test) Regression → Block 1/2 → Statistics: R² ESSENTIAL
change
Save Residuals Regression → Save → Unstandardized ESSENTIAL
Residuals + Predicted
Residual Plot Graph → Chart Builder → Scatter (predicted vs ESSENTIAL
residuals)
VIF / Tolerance Regression → Statistics → Collinearity HIGH YIELD
diagnostics
Create Lag Variables Transform → Create Time Series → Lag ESSENTIAL
Create Difference Variables Transform → Create Time Series → Difference ESSENTIAL
Compute New Variable Transform → Compute Variable ESSENTIAL
Time Series Plot Analyze → Forecasting → Sequence Charts HIGH YIELD
Correlations Analyze → Correlate → Bivariate HIGH YIELD
Rania El Ghalbzouri Page 3
MBA Bridging Programme
, QRM Exam Survival Masterclass
STEP 2: PROFESSOR PATTERN ANALYSIS
Your professor has very consistent questioning patterns. Below is every recurring topic with: why it is asked,
how the question looks, and the exact answer structure that earns full marks.
2.1 Coefficient Interpretation
Why asked: This is the single most frequent question. The professor tests whether you can translate
regression output into real-world meaning with correct units, direction, significance, and ceteris paribus.
How it looks: "Discuss the effect of X on Y based on your fitted model" or "Interpret all effects in the model."
Full-Marks Answer Templates
TEMPLATE: Normal Variable
If [X] increases by one [unit], [Y] increases/decreases on average by [b] [units], ceteris paribus. This effect is
significant/not significant (t = [value], p-value = [value]).
TEMPLATE: Dummy Variable
Compared to [reference category], [category name] has on average a [b] [units] higher/lower [Y], ceteris
paribus. This effect is significant/not significant (t = [value], p-value = [value]).
TEMPLATE: Log-Level — ln(X) in model, Y level
If [X] increases by 1%, [Y] changes on average by β/100 [units], ceteris paribus. This effect is significant (t =
[value], p = [value]).
TEMPLATE: Log-Log — ln(Y) and ln(X)
If [X] increases by 1%, [Y] increases/decreases on average by approximately β%, ceteris paribus. This effect is
significant (t = [value], p = [value]).
TEMPLATE: Level-Log — Y level, ln(X)
If [X] increases by 1%, [Y] changes on average by β/100 [units], ceteris paribus.
TEMPLATE: Dummy in Log Model — ln(Y)
Compared to [ref], [category] has on average approximately (β × 100)% higher/lower [Y], ceteris paribus. For
large β, use (eβ − 1) × 100%.
Rania El Ghalbzouri Page 4
MBA Bridging Programme
QRM EXAM
SURVIVAL MASTERCLASS
The Complete Exam-Solving Handbook
Prof. Goedele Dierckx
Quantitative Research Methods 2025–2026
Exam Duration: 2.5 hours │ Tools: SPSS + Excel
MLR │ Time Series
Based on reverse-engineering of all available past exams and official solutions
Rania El Ghalbzouri Page 1
MBA Bridging Programme
, QRM Exam Survival Masterclass
STEP 1: EXAM ANALYSIS
Analysis of all provided exams reveals clear patterns. The following frequency table ranks every recurring
question type by how often it appeared and its probability of appearing on your exam.
1.1 Question Type Frequency Table
Question Type Freq Importance Exam Prob.
Coefficient interpretation (incl. dummies, logs) 12+ ESSENTIAL 100%
Write theoretical model equation + assumptions 8+ ESSENTIAL 100%
Joint F-test (R² change) 7+ ESSENTIAL 95%
Omitted variable bias analysis 6+ ESSENTIAL 90%
Residual plot interpretation 6+ ESSENTIAL 90%
Dummy variable creation + interpretation 8+ ESSENTIAL 100%
Interaction effect interpretation 6+ ESSENTIAL 85%
VIF / Tolerance / QMC check 5+ HIGH YIELD 80%
Auxiliary regression model 5+ HIGH YIELD 80%
White test (write auxiliary model) 3+ HIGH YIELD 70%
Log-level / log-log interpretation 6+ ESSENTIAL 90%
Dickey–Fuller test for stationarity 6+ ESSENTIAL 95%
AR(1) model + multiplier effects 4+ HIGH YIELD 80%
Distributed lag model + lag elimination 4+ HIGH YIELD 80%
Autocorrelation test (LM / Breusch–Godfrey) 5+ ESSENTIAL 85%
Short-run vs long-run effects 4+ HIGH YIELD 80%
R² / Adjusted R² interpretation 5+ HIGH YIELD 75%
Residual calculation for specific observation 3+ NICE TO KNOW 50%
Spurious regression concept 2+ NICE TO KNOW 40%
Stationarity of first differences 4+ HIGH YIELD 75%
Rania El Ghalbzouri Page 2
MBA Bridging Programme
, QRM Exam Survival Masterclass
1.2 Recurring Formula Summary
Formula Used For Priority
F = (R²full − R²restricted) / J ÷ (1 − R²full) / (n − k − 1) Joint F-test ESSENTIAL
F = (SSER − SSE) / J ÷ SSE / (n − k − 1) Joint F-test (alt) ESSENTIAL
Total multiplier = β / (1 − γ) AR(1) long-run effect ESSENTIAL
Tolerance = 1 − R²aux Multicollinearity check HIGH YIELD
VIF = 1 / Tolerance Multicollinearity check HIGH YIELD
Test stat = nR²aux (χ², df = 1) LM autocorrelation test ESSENTIAL
Test stat = nR²aux (χ², df = p) White test HIGH YIELD
Log interpretation: β/100 for 1% change Log-level model ESSENTIAL
Dummy log: approx (eβ − 1) × 100% Log-dummy effect HIGH YIELD
Critical value DF with trend: t* = −3.41 Dickey–Fuller (5%) ESSENTIAL
Critical value DF no trend: t* = −2.86 Dickey–Fuller (5%) ESSENTIAL
1.3 Recurring SPSS Procedures
SPSS Task Menu Path Priority
Linear Regression Analyze → Regression → Linear ESSENTIAL
R² Change (Joint F-test) Regression → Block 1/2 → Statistics: R² ESSENTIAL
change
Save Residuals Regression → Save → Unstandardized ESSENTIAL
Residuals + Predicted
Residual Plot Graph → Chart Builder → Scatter (predicted vs ESSENTIAL
residuals)
VIF / Tolerance Regression → Statistics → Collinearity HIGH YIELD
diagnostics
Create Lag Variables Transform → Create Time Series → Lag ESSENTIAL
Create Difference Variables Transform → Create Time Series → Difference ESSENTIAL
Compute New Variable Transform → Compute Variable ESSENTIAL
Time Series Plot Analyze → Forecasting → Sequence Charts HIGH YIELD
Correlations Analyze → Correlate → Bivariate HIGH YIELD
Rania El Ghalbzouri Page 3
MBA Bridging Programme
, QRM Exam Survival Masterclass
STEP 2: PROFESSOR PATTERN ANALYSIS
Your professor has very consistent questioning patterns. Below is every recurring topic with: why it is asked,
how the question looks, and the exact answer structure that earns full marks.
2.1 Coefficient Interpretation
Why asked: This is the single most frequent question. The professor tests whether you can translate
regression output into real-world meaning with correct units, direction, significance, and ceteris paribus.
How it looks: "Discuss the effect of X on Y based on your fitted model" or "Interpret all effects in the model."
Full-Marks Answer Templates
TEMPLATE: Normal Variable
If [X] increases by one [unit], [Y] increases/decreases on average by [b] [units], ceteris paribus. This effect is
significant/not significant (t = [value], p-value = [value]).
TEMPLATE: Dummy Variable
Compared to [reference category], [category name] has on average a [b] [units] higher/lower [Y], ceteris
paribus. This effect is significant/not significant (t = [value], p-value = [value]).
TEMPLATE: Log-Level — ln(X) in model, Y level
If [X] increases by 1%, [Y] changes on average by β/100 [units], ceteris paribus. This effect is significant (t =
[value], p = [value]).
TEMPLATE: Log-Log — ln(Y) and ln(X)
If [X] increases by 1%, [Y] increases/decreases on average by approximately β%, ceteris paribus. This effect is
significant (t = [value], p = [value]).
TEMPLATE: Level-Log — Y level, ln(X)
If [X] increases by 1%, [Y] changes on average by β/100 [units], ceteris paribus.
TEMPLATE: Dummy in Log Model — ln(Y)
Compared to [ref], [category] has on average approximately (β × 100)% higher/lower [Y], ceteris paribus. For
large β, use (eβ − 1) × 100%.
Rania El Ghalbzouri Page 4
MBA Bridging Programme
