CMN 548 TEST 4| ACTUAL EXAM QUESTIONS AND
CORRECT ANSWERS WITH RATIONALES
GRADED A+ LATEST
CMN 548 — Test 4
1
A public health team wants to evaluate whether a new anti-smoking PSA reduces
intentions to smoke among adolescents. They randomly assign schools to either
show the PSA (treatment) or not (control) and measure intentions before and after.
Which design best describes this study?
A. Cross-sectional survey
B. Pretest-posttest randomized control trial (cluster randomized)
C. Interrupted time series
D. Quasi-experimental nonequivalent groups
Answer: B
Rationale: Schools were randomly assigned to treatment or control (randomization
at the cluster level), and intentions are measured before and after — this is a
pretest-posttest randomized control trial with clustering. It’s not cross-sectional
(which measures one time), not interrupted time series (requires multiple time
points), and not quasi-experimental since random assignment was used.
2
Researchers conducting a content analysis code newspapers for the presence of
frames (e.g., economic, moral, legal). They calculate Krippendorff’s alpha for
inter-coder reliability and obtain 0.55. What is the best interpretation?
A. Excellent reliability — proceed without changes.
B. Acceptable reliability for exploratory research.
,C. Poor reliability — revise the codebook and retrain coders.
D. Reliability is irrelevant for content analysis.
Answer: C
Rationale: Krippendorff’s alpha values above ~0.80 are generally considered
reliable; 0.55 is low. The appropriate action is to revise the codebook, clarify
coding rules, and retrain coders to improve reliability. It’s not excellent or
acceptable; reliability is essential.
3
Which sampling technique is most likely to produce a sample representative of the
population's age distribution if age is known for the population?
A. Simple random sampling
B. Snowball sampling
C. Stratified random sampling (by age group)
D. Convenience sampling
Answer: C
Rationale: Stratified sampling ensures representation across known subgroups
(age strata) by sampling within each stratum proportionally. Simple random
sampling can by chance under/over-represent age groups; snowball and
convenience are non-probability and not representative.
4
A study finds a statistically significant correlation of r = 0.32 (p < .01) between
social media use and depressive symptoms. Which statement is correct?
A. Social media use causes depression.
B. The correlation indicates a strong relationship.
C. The correlation indicates a moderate relationship; causality cannot be inferred.
,D. The correlation is negligible and unimportant.
Answer: C
Rationale: An r of 0.32 is a small-to-moderate effect size. Correlation does not
imply causation — without experimental or longitudinal evidence, we can't claim
causality.
5
A health campaign uses the Extended Parallel Process Model (EPPM) to design
messages. Which combination of message elements is predicted to produce danger
control (adaptive action) rather than fear control (defensive avoidance)?
A. High perceived threat, low perceived efficacy
B. Low perceived threat, high perceived efficacy
C. High perceived threat, high perceived efficacy
D. Low perceived threat, low perceived efficacy
Answer: C
Rationale: EPPM predicts danger control when both perceived threat (severity +
susceptibility) and perceived efficacy (self + response efficacy) are high. High
threat with low efficacy leads to fear control.
6 (Scenario)
A nonprofit runs A/B tests on two versions of a donation landing page. Version A
shows faces of beneficiaries; Version B shows statistics about impact. Over one
week, Version A converts 3.1% (310/10,000) and Version B converts 2.6%
(260/10,000). Which statistical test is most appropriate to compare conversion
rates and why?
A. Independent-samples t-test — because you compare two means.
B. Chi-square test of independence — because you compare categorical outcomes
between groups.
, C. Paired t-test — because observations are paired.
D. ANOVA — because it handles proportions.
Answer: B
Rationale: Conversion (converted / not converted) is categorical. Comparing
proportions across two independent groups is appropriately tested with a chi-
square test (or a two-proportion z-test). t-tests compare continuous means; paired t-
test requires paired data; ANOVA is for comparing means across 3+ groups.
7
In qualitative interviews, a researcher intentionally seeks participants who
represent extreme perspectives to understand rare experiences. What is this
sampling strategy called?
A. Purposive sampling — maximum variation/extreme case sampling
B. Quota sampling
C. Snowball sampling
D. Convenience sampling
Answer: A
Rationale: Purposive sampling includes strategies like maximum variation and
extreme case sampling to deliberately select participants with particular
characteristics or perspectives. Quota and snowball are different nonprobability
approaches.
8
Which validity type addresses whether a survey measure actually captures the
theoretical construct it intends to measure?
A. Face validity
B. Construct validity
CORRECT ANSWERS WITH RATIONALES
GRADED A+ LATEST
CMN 548 — Test 4
1
A public health team wants to evaluate whether a new anti-smoking PSA reduces
intentions to smoke among adolescents. They randomly assign schools to either
show the PSA (treatment) or not (control) and measure intentions before and after.
Which design best describes this study?
A. Cross-sectional survey
B. Pretest-posttest randomized control trial (cluster randomized)
C. Interrupted time series
D. Quasi-experimental nonequivalent groups
Answer: B
Rationale: Schools were randomly assigned to treatment or control (randomization
at the cluster level), and intentions are measured before and after — this is a
pretest-posttest randomized control trial with clustering. It’s not cross-sectional
(which measures one time), not interrupted time series (requires multiple time
points), and not quasi-experimental since random assignment was used.
2
Researchers conducting a content analysis code newspapers for the presence of
frames (e.g., economic, moral, legal). They calculate Krippendorff’s alpha for
inter-coder reliability and obtain 0.55. What is the best interpretation?
A. Excellent reliability — proceed without changes.
B. Acceptable reliability for exploratory research.
,C. Poor reliability — revise the codebook and retrain coders.
D. Reliability is irrelevant for content analysis.
Answer: C
Rationale: Krippendorff’s alpha values above ~0.80 are generally considered
reliable; 0.55 is low. The appropriate action is to revise the codebook, clarify
coding rules, and retrain coders to improve reliability. It’s not excellent or
acceptable; reliability is essential.
3
Which sampling technique is most likely to produce a sample representative of the
population's age distribution if age is known for the population?
A. Simple random sampling
B. Snowball sampling
C. Stratified random sampling (by age group)
D. Convenience sampling
Answer: C
Rationale: Stratified sampling ensures representation across known subgroups
(age strata) by sampling within each stratum proportionally. Simple random
sampling can by chance under/over-represent age groups; snowball and
convenience are non-probability and not representative.
4
A study finds a statistically significant correlation of r = 0.32 (p < .01) between
social media use and depressive symptoms. Which statement is correct?
A. Social media use causes depression.
B. The correlation indicates a strong relationship.
C. The correlation indicates a moderate relationship; causality cannot be inferred.
,D. The correlation is negligible and unimportant.
Answer: C
Rationale: An r of 0.32 is a small-to-moderate effect size. Correlation does not
imply causation — without experimental or longitudinal evidence, we can't claim
causality.
5
A health campaign uses the Extended Parallel Process Model (EPPM) to design
messages. Which combination of message elements is predicted to produce danger
control (adaptive action) rather than fear control (defensive avoidance)?
A. High perceived threat, low perceived efficacy
B. Low perceived threat, high perceived efficacy
C. High perceived threat, high perceived efficacy
D. Low perceived threat, low perceived efficacy
Answer: C
Rationale: EPPM predicts danger control when both perceived threat (severity +
susceptibility) and perceived efficacy (self + response efficacy) are high. High
threat with low efficacy leads to fear control.
6 (Scenario)
A nonprofit runs A/B tests on two versions of a donation landing page. Version A
shows faces of beneficiaries; Version B shows statistics about impact. Over one
week, Version A converts 3.1% (310/10,000) and Version B converts 2.6%
(260/10,000). Which statistical test is most appropriate to compare conversion
rates and why?
A. Independent-samples t-test — because you compare two means.
B. Chi-square test of independence — because you compare categorical outcomes
between groups.
, C. Paired t-test — because observations are paired.
D. ANOVA — because it handles proportions.
Answer: B
Rationale: Conversion (converted / not converted) is categorical. Comparing
proportions across two independent groups is appropriately tested with a chi-
square test (or a two-proportion z-test). t-tests compare continuous means; paired t-
test requires paired data; ANOVA is for comparing means across 3+ groups.
7
In qualitative interviews, a researcher intentionally seeks participants who
represent extreme perspectives to understand rare experiences. What is this
sampling strategy called?
A. Purposive sampling — maximum variation/extreme case sampling
B. Quota sampling
C. Snowball sampling
D. Convenience sampling
Answer: A
Rationale: Purposive sampling includes strategies like maximum variation and
extreme case sampling to deliberately select participants with particular
characteristics or perspectives. Quota and snowball are different nonprobability
approaches.
8
Which validity type addresses whether a survey measure actually captures the
theoretical construct it intends to measure?
A. Face validity
B. Construct validity
