Difficulty: Hard
A sample has 40 units from a population of 400. What is its sampling fraction?
A 4%
B 40%
C 90%
D 10%
Answer: D. 10%. The sampling fraction is 40 divided by 400, or 10%.
A survey sends 250 forms and receives 200 usable replies. What is the response rate?
A 80%
B 50%
C 20%
D 90%
Answer: A. 80%. The response rate is 200 divided by 250, multiplied by 100.
A stratum is 25% of the population. A proportional sample has 200 units. How many come from it?
A 25
B 50
C 75
D 100
Answer: B. 50. Proportional allocation assigns 25% of 200, which is 50 units.
A list has 1,000 units and a sample of 100 is needed systematically. What is the interval?
A 5
B 20
C 10
D 100
Answer: C. 10. The interval is the frame size divided by sample size: 1,000 divided by 100.
A sample includes 50 people and 12 do not respond. What is the response rate?
A 24%
B 88%
C 62%
D 76%
Answer: D. 76%. Thirty-eight people respond; 38 divided by 50 equals 76%.
Which problem can remain even in a census?
A Non-sampling errors
B Sampling error from selection
C A sampling fraction below 100%
D A missing sample stratum
Answer: A. Non-sampling errors. A census avoids sampling error but can still have nonresponse, measurement, coverage, and processing errors.
A sample estimate varies because only some units were selected. What error is this?
A Measurement error
B Sampling error
C Interviewer error
D Data-entry error
Answer: B. Sampling error. Sampling error reflects variation from observing a sample rather than the entire population.
A 100-person sample misses an entire remote district. Which problem is most direct?
A Random sampling variation
B A seasonal price effect
C Coverage error
D A high response rate
Answer: C. Coverage error. The sample frame or design failed to cover part of the target population.
Why allocate a stratified sample proportionally?
A To choose the easiest group
B To avoid all random selection
C To make every stratum equal
D To match population group shares
Answer: D. To match population group shares. Proportional allocation gives each stratum a sample share corresponding to its population share.
A survey is emailed only to volunteers. What is the main limitation?
A Self-selection bias
B A complete census design
C Known equal selection chances
D Automatic national representation
Answer: A. Self-selection bias. Volunteers may differ from nonparticipants, so results may not represent the target population.
What is the value of a random start in a list sample?
A It converts samples to censuses
B It reduces arbitrary selection bias
C It prevents all nonresponse
D It guarantees subgroup balance
Answer: B. It reduces arbitrary selection bias. A random start gives the systematic procedure a chance-based beginning.
Why follow up nonrespondents?
A To change the target population
B To guarantee perfect answers
C Improve response, assess bias
D To bypass consent
Answer: C. Improve response, assess bias. Follow-up can increase response and help assess whether nonresponse affects the results.
A questionnaire gives the same result when repeated under similar conditions. What quality is this?
A Coverage
B External validity
C Sampling fraction
D Reliability
Answer: D. Reliability. Reliability concerns the consistency of measurement under similar conditions.
A question measures the concept it is intended to measure. What quality is this?
A Validity
B Response rate
C Nonresponse
D Randomisation
Answer: A. Validity. Validity concerns whether the instrument measures the intended concept.
Why pretest a translated questionnaire?
A Calculate national average
B To check meaning and clarity
C To select the sample frame
D To remove every cultural difference
Answer: B. To check meaning and clarity. Pretesting can reveal wording that is unclear or changes meaning across languages.
A sample survey studies urban and rural households separately. Which design fits?
A Convenience sampling
B Snowball sampling
C Stratified sampling
D A full census
Answer: C. Stratified sampling. Urban and rural areas are strata, with units sampled within each group.
A researcher randomly chooses 12 villages and surveys all households there. What is this?
A Systematic individual sampling
B Purposive sampling
C Simple random household sampling
D Cluster sampling
Answer: D. Cluster sampling. The researcher selects village clusters and observes households within them.
A report compares two years but changes its income definition. What is the risk?
A Measures may not compare
B The sample becomes a census
C The response rate doubles
D The population definition is fixed
Answer: A. Measures may not compare. Changing definitions can make apparent differences reflect measurement changes rather than real change.
A sampling frame contains duplicate household entries. What can result?
A Every household counted once
B Some households get extra chances
C Nonresponse falls automatically
D The sample becomes stratified
Answer: B. Some households get extra chances. Duplicates may give some units multiple chances and distort selection probabilities.
How should personally identifying data be handled?
A Publish openly by default
B Share with unrelated firms
C Store securely and limit access
D Keep no consent record
Answer: C. Store securely and limit access. Personal data should be stored securely and accessed only for legitimate study purposes.
A published survey has no method notes. What is the best next step?
A Assume the sample was random
B Treat all estimates as exact
C Compare only its headline number
D Find documentation first
Answer: D. Find documentation first. Method documentation is needed to assess sample design, definitions, and limitations.
Why might a smaller probability sample beat a larger volunteer sample?
A It may represent population better
B It always has zero error
C Volunteers answer incorrectly
D Sample size has no effect
Answer: A. It may represent population better. A well-designed probability sample can be more representative than a larger self-selected group.
A researcher asks farmers about last year’s crop output. What can reduce recall error?
A Ask a leading question
B Use records and dates
C Remove the reference dates
D Use only memory prompts
Answer: B. Use records and dates. Farm records and a clearly defined reference period can improve recall accuracy.
Why distinguish census from sample-survey results?
A Census has zero errors
B Samples include every unit
C Their coverage and errors differ
D Both produce identical information
Answer: C. Their coverage and errors differ. A census covers all units but can have nonsampling errors; a sample adds sampling uncertainty.
What should a report state about its sample?
A Only the final percentage
B Respondents’ personal names
C A claim of perfect accuracy
D Selection method and limitations
Answer: D. Selection method and limitations. Reporting the sample design and limitations helps readers judge how far results can be generalised.