MODULE 4 QUESTIONS
SHORTNOTE
Level 2: Understanding
At this level, questions focus on explaining ideas or concepts.
1. Explain the core difference between an "artificial dichotomy" and a "natural dichotomy" as
they relate to correlational studies.
2. Describe the conceptual foundation of the Chi-square test in terms of comparing "Observed
Frequencies" and "Expected Frequencies".
3. Summarize the rationale for using non-parametric tests, also known as "distribution-free
tests," in psychological research.
4. Discuss the concept of "interaction effects" in Two-Way ANOVA and why they are often
more meaningful than main effects.
5. Interpret what a "monotonic relationship" signifies in the context of Spearman’s Rank
Correlation.
Level 3: Applying
These questions require using information in new but similar situations.
1. Identify the appropriate correlational technique for a study measuring the relationship
between a student's "pass/fail" status (artificially split at 50%) and their continuous final
exam score.
2. Calculate the degrees of freedom (df) for a Chi-square Test of Independence involving a
contingency table with 3 rows (teaching methods) and 2 columns (gender).
3. Determine the cell count and design type for a Two-Way ANOVA study using 3 levels of
"Teaching Method" and 2 levels of "Prior Knowledge".
4. Demonstrate the first three steps of calculating Spearman’s Rank Correlation for a small
dataset of ranked preferences.
5. Apply the concept of a "Balanced Design" to explain how a researcher should organize
participant groups in a 2x2 factorial experiment.
Level 4: Analyzing
Questions here focus on breaking information into parts to explore understandings and relationships.
1. Differentiate between the assumptions of Pearson’s Correlation and those required for
Biserial Correlation.
2. Analyze the partitioning of variance in a Two-Way ANOVA, explaining the four specific
components that make up total variability.
3. Compare and contrast the Spearman Rank Correlation with the Point-Biserial Correlation in
terms of data types and normality assumptions.
4. Examine the limitations of the Biserial correlation coefficient (r_{bis}), specifically regarding
its range and use in regression equations.
5. Contrast the Sign Test and the Wilcoxon Test (from general non-parametric principles) to
determine which is more powerful for paired data.
Level 5: Evaluating
At this level, questions require justifying a stand or decision.
1. Critique the use of a parametric t-test when a researcher discovers their data has a highly
skewed distribution and small sample size.
2. Justify the selection of Yates’ Correction in a Chi-square analysis involving a 2x2 table.
3. Evaluate the impact of violating the "Homogeneity of Variance" assumption on the validity
of a Two-Way ANOVA F-statistic.
4. Assess the advantages and disadvantages of using Spearman’s Rank Correlation over
Pearson’s Correlation when data contains extreme outliers.
5. Appraise the decision to use a Biserial Correlation instead of a Point-Biserial Correlation for a
variable like "Social Adjustment".
Level 6: Creating
These questions involve putting parts together to form a whole or proposing alternative solutions.
1. Design a 3x3 Factorial study in educational psychology, specifying the factors, levels, and a
potential interaction effect to investigate.
2. Formulate a set of Null and Alternative Hypotheses for a Chi-square Test of Independence
between "Therapy Type" and "Recovery Outcome".
3. Develop a step-by-step guideline for a junior researcher on how to choose between
Spearman, Biserial, and Point-Biserial techniques based on data characteristics.
4. Propose a research scenario where an "Interaction Model" (non-parallel lines) would be
more realistic than an "Additive Model".
5. Construct a mock contingency table for a "Median Test" between two independent groups
and describe how you would conclude if they differ significantly.
ESSAY TYPE
Level 2: Understanding (Comprehension)
Focus: Explaining ideas or concepts in your own words.
1. Differentiate Parametric and Non-parametric Tests: Explain the fundamental differences
between parametric and non-parametric (distribution-free) tests. Discuss why non-
parametric tests are often preferred when dealing with small sample sizes or non-normal
distributions.
2. Define Main Effects and Interactions: In the context of Factorial ANOVA, define "main
effects" and "interaction effects". Explain why it is critical to interpret main effects with
caution when a significant interaction is present.
3. Contrast Biserial and Point-Biserial Correlation: Compare the use cases for Biserial and
Point-Biserial correlations. Specifically, explain the difference between a "natural dichotomy"
and an "artificial dichotomy" as they relate to these techniques.
4. Explain the Logic of Chi-Square: Describe the conceptual foundation of the Chi-Square test,
focusing on the relationship between "Observed Frequencies" and "Expected Frequencies".
What does it mean for a result to be statistically significant in this context?.
5. Summarize Resampling Approaches: Explain the general approach of "resampling tests,"
including bootstrapping and randomization tests. How do these methods differ from
traditional parametric estimation?.
Level 3: Applying
Focus: Using information in another familiar situation.
1. Selecting a Correlation Technique: You are researching the relationship between student IQ
(continuous) and whether they passed a specific exam (artificial dichotomy). Determine
which correlational technique is most appropriate and justify your choice based on the
nature of the variables.
2. Applying Non-parametric Tests: A researcher has a very small sample size (N=10) with highly
skewed data and wants to compare two independent groups. Which specific non-parametric
test should they apply, and what assumptions are they avoiding by doing so?.
3. Chi-Square Implementation: A hospital wants to know if a new drug's success rate differs
between male and female patients. Explain how a 2 \times 2 contingency table would be set
up for this study and how the expected frequencies would be calculated.
4. ANOVA Cell Diagramming: Using the 2 \times 2 design example of alcohol and expectations
on aggression, describe how to organize the data into a cell diagram. Explain what the row
and column means represent in this specific layout.
5. Calculating Spearman’s Rank: If you are given a dataset consisting of rankings for two
variables rather than raw scores, describe the step-by-step process of applying Spearman’s
Rank Correlation.
Level 4: Analyzing
Focus: Breaking information into parts to explore understandings and relationships.
1. Comparing ANOVA vs. Multiple T-Tests: Analyze the advantages of using a Two-Way ANOVA
over conducting multiple separate one-way analyses or t-tests. How does the Two-Way
ANOVA help in understanding the multidimensional nature of human behavior?
2. Robustness Analysis: Evaluate the conclusion that non-parametric tests like the Mann-
Whitney U are more powerful than the t-test when underlying assumptions are violated.
Under what specific conditions does the t-test lose its robustness?.
3. Interaction Patterns: Analyze a hypothetical interaction plot where two lines cross. What
does this "crossover interaction" tell the researcher about the relationship between the
independent variables and the dependent variable?.
4. Chi-Square Limitations: Discuss the "common errors" in interpreting Chi-Square results, such
as the confusion between significance and strength. Why is reporting an effect size (like Phi
or Cramer’s V) necessary for a complete analysis?.
5. Biserial Correlation Assumptions: Analyze the normality and variance assumptions required
for Biserial correlation. What happens to the validity of the result if the underlying
continuous distribution is not normal?.
Level 5: Evaluating
Focus: Justifying a stand or decision.
1. Critique the Choice of Statistics: A researcher uses a Pearson r to correlate rank-ordered
data. Critique this decision based on the assumptions of the Pearson coefficient and suggest
a more appropriate "Alternative Correlational Technique".
2. Assessing Parametric vs. Non-parametric Value: Evaluate the statement: "Parametric tests
are always superior to non-parametric tests because they use more information from the
data." Support or refute this using evidence regarding Type I and Type II errors.
3. Interpreting Non-significant ANOVA: If a Two-Way ANOVA yields a significant interaction but
non-significant main effects, evaluate how the researcher should proceed with their
conclusions. Is it still meaningful to report the results?.
4. Chi-Square for Ordinal Data: Is it appropriate to use a standard Chi-Square test for data that
has a natural order (ordinal data)?. Evaluate the use of the Mantel-Haenszel statistic or other
"measures of agreement" as alternatives.
5. Ethical Considerations in Data Dichotomization: Evaluate the practice of "artificially
reducing" a continuous variable to a dichotomy (e.g., "Pass/Fail"). What are the potential
statistical and practical risks of losing data nuance in this way?.
Level 6: Creating (Synthesis)
Focus: Generating new ideas, products, or ways of viewing things.
1. Design a Factorial Study: Develop a 3 \times 2 factorial research design to study the effects
of "Teaching Method" and "Student Motivation Level" on "Exam Performance." Clearly
identify the factors, levels, and a potential interaction you might expect to find.
2. Develop a Statistical Decision Tree: Create a comprehensive flowchart or decision tree that a
researcher can use to choose between a t-test, Mann-Whitney U, Pearson r, Spearman’s
Rank, and Chi-Square based on their data type and assumptions.
3. Propose a Resampling Experiment: Design a study that uses bootstrapping to determine the
confidence interval for a correlation coefficient in a situation where the population
distribution is unknown.
4. Synthesize a Comprehensive Analysis Plan: Imagine you are analyzing the effectiveness of a
mental health intervention using categorical outcomes (Improved/Not Improved) and
continuous self-report scores. Construct a plan that utilizes both Chi-Square and Point-
Biserial correlation to provide a holistic view of the data.
5. Formulate a Hybrid Theory: Propose a "best practices" guide for researchers that integrates
both parametric and non-parametric approaches. When should both be run simultaneously
to "triangulate" findings?.