, Question 1
A graduate team wants to estimate the causal effect of a new pedagogy on
learning. They randomly assign students within classrooms but worry about
spillover between treated and control peers. Which design element best
mitigates contamination while preserving random assignment?
A. Individually randomized trial with intention-to-treat analysis
B. Cluster randomized trial with randomization at the classroom level
C. Stepped-wedge design with delayed implementation
D. Regression discontinuity using a cutoff score
Correct Answer: B - Cluster randomized trial with
randomization at the classroom level
RATIONALE
Spillover within classrooms violates the no-interference assumption of
individual randomization; randomizing intact clusters (classrooms)
contains contamination and preserves causal identification. ITT alone
does not address interference, stepped-wedge still mixes clusters over
time, and regression discontinuity does not randomize.
Question 2
A scale intended to measure 'research self-efficacy' yields a Cronbach's alpha
of 0.93, but factor analysis shows two distinct dimensions. Which conclusion is
most defensible?
A. High alpha confirms unidimensionality and the scale is ready for use.
B. High alpha indicates strong reliability; dimensionality must be
addressed separately.
C. The scale should be discarded because alpha exceeds 0.90.
D. Alpha is irrelevant when factor analysis is available.
Correct Answer: B - High alpha indicates strong reliability;
dimensionality must be addressed separately.
Page 2
, RATIONALE
Alpha estimates internal consistency, not dimensionality; a
multidimensional scale can still produce high alpha. Factor analysis
reveals the underlying structure, so the scale may need subscale
scoring. Discarding a reliable scale is unwarranted, and alpha remains
a useful reliability index alongside factor analysis.
Question 3
In a two-stage cluster sample with unequal cluster sizes, which weighting
strategy yields an approximately unbiased design-based estimate of the
population mean?
A. Equal weights for all respondents
B. Inverse-probability weights at each stage
C. Post-stratification weights only
D. Weights proportional to cluster size alone
Correct Answer: B - Inverse-probability weights at each stage
RATIONALE
Design-based unbiasedness requires each unit's weight to equal the
inverse of its selection probability, compounded across stages. Equal
weights ignore selection probabilities, post-stratification alone does
not correct selection bias, and cluster-size-only weights omit
within-cluster selection probabilities.
Question 4
A researcher reports a 95% confidence interval for a mean difference of [0.2,
1.8]. Which interpretation is correct under frequentist theory?
A. There is a 95% probability the true difference lies in this interval.
B. 95% of future sample differences will fall in this interval.
C. The procedure yields intervals covering the true value in 95% of
repeated samples.
D. The true difference is definitely between 0.2 and 1.8.
Page 3
, Correct Answer: C - The procedure yields intervals covering the
true value in 95% of repeated samples.
RATIONALE
Frequentist coverage refers to the long-run performance of the
interval-construction procedure, not the probability of a fixed
parameter falling in a realized interval. Options A and D assign
probability to a fixed parameter, and B confuses the interval with a
prediction interval.
Question 5
A researcher uses grounded theory but reports only themes with no constant
comparison or theoretical sampling. Which methodological critique is most
warranted?
A. The study is actually phenomenology, not grounded theory.
B. The study lacks core grounded theory procedures and cannot claim
theory generation.
C. The study is valid if saturation was achieved.
D. The study should have used ethnography instead.
Correct Answer: B - The study lacks core grounded theory
procedures and cannot claim theory generation.
RATIONALE
Grounded theory requires iterative constant comparison and
theoretical sampling to generate theory; thematic analysis alone does
not meet these criteria. It is not necessarily phenomenology or
ethnography, and saturation without those procedures does not confer
grounded-theory status.
Page 4
A graduate team wants to estimate the causal effect of a new pedagogy on
learning. They randomly assign students within classrooms but worry about
spillover between treated and control peers. Which design element best
mitigates contamination while preserving random assignment?
A. Individually randomized trial with intention-to-treat analysis
B. Cluster randomized trial with randomization at the classroom level
C. Stepped-wedge design with delayed implementation
D. Regression discontinuity using a cutoff score
Correct Answer: B - Cluster randomized trial with
randomization at the classroom level
RATIONALE
Spillover within classrooms violates the no-interference assumption of
individual randomization; randomizing intact clusters (classrooms)
contains contamination and preserves causal identification. ITT alone
does not address interference, stepped-wedge still mixes clusters over
time, and regression discontinuity does not randomize.
Question 2
A scale intended to measure 'research self-efficacy' yields a Cronbach's alpha
of 0.93, but factor analysis shows two distinct dimensions. Which conclusion is
most defensible?
A. High alpha confirms unidimensionality and the scale is ready for use.
B. High alpha indicates strong reliability; dimensionality must be
addressed separately.
C. The scale should be discarded because alpha exceeds 0.90.
D. Alpha is irrelevant when factor analysis is available.
Correct Answer: B - High alpha indicates strong reliability;
dimensionality must be addressed separately.
Page 2
, RATIONALE
Alpha estimates internal consistency, not dimensionality; a
multidimensional scale can still produce high alpha. Factor analysis
reveals the underlying structure, so the scale may need subscale
scoring. Discarding a reliable scale is unwarranted, and alpha remains
a useful reliability index alongside factor analysis.
Question 3
In a two-stage cluster sample with unequal cluster sizes, which weighting
strategy yields an approximately unbiased design-based estimate of the
population mean?
A. Equal weights for all respondents
B. Inverse-probability weights at each stage
C. Post-stratification weights only
D. Weights proportional to cluster size alone
Correct Answer: B - Inverse-probability weights at each stage
RATIONALE
Design-based unbiasedness requires each unit's weight to equal the
inverse of its selection probability, compounded across stages. Equal
weights ignore selection probabilities, post-stratification alone does
not correct selection bias, and cluster-size-only weights omit
within-cluster selection probabilities.
Question 4
A researcher reports a 95% confidence interval for a mean difference of [0.2,
1.8]. Which interpretation is correct under frequentist theory?
A. There is a 95% probability the true difference lies in this interval.
B. 95% of future sample differences will fall in this interval.
C. The procedure yields intervals covering the true value in 95% of
repeated samples.
D. The true difference is definitely between 0.2 and 1.8.
Page 3
, Correct Answer: C - The procedure yields intervals covering the
true value in 95% of repeated samples.
RATIONALE
Frequentist coverage refers to the long-run performance of the
interval-construction procedure, not the probability of a fixed
parameter falling in a realized interval. Options A and D assign
probability to a fixed parameter, and B confuses the interval with a
prediction interval.
Question 5
A researcher uses grounded theory but reports only themes with no constant
comparison or theoretical sampling. Which methodological critique is most
warranted?
A. The study is actually phenomenology, not grounded theory.
B. The study lacks core grounded theory procedures and cannot claim
theory generation.
C. The study is valid if saturation was achieved.
D. The study should have used ethnography instead.
Correct Answer: B - The study lacks core grounded theory
procedures and cannot claim theory generation.
RATIONALE
Grounded theory requires iterative constant comparison and
theoretical sampling to generate theory; thematic analysis alone does
not meet these criteria. It is not necessarily phenomenology or
ethnography, and saturation without those procedures does not confer
grounded-theory status.
Page 4
