A researcher wants to estimate what percentage of residents in a large city support a new transit policy. Which sampling strategy is most appropriate, and why?
APurposive sampling, to ensure the most knowledgeable and engaged residents are included
BSnowball sampling, because transit policy affects interconnected social networks
CProbability sampling (e.g., random or stratified), because it enables statistically valid inference about the full population
DTheoretical sampling, continuing until no new viewpoints emerge
The research goal is a population-level proportion estimate, which requires a probability sample. Only when every member of the population has a known, non-zero chance of selection can the sample's results be generalized with calculable margins of error. Purposive sampling (A) introduces systematic bias by selecting on a criterion; snowball sampling (B) over-represents connected clusters; theoretical sampling (D) is a grounded-theory method for conceptual development, not population estimation. Matching sampling strategy to research question is the central principle.
Question 2 Multiple Choice
A researcher studying survival experiences among undocumented immigrants begins with contacts from an advocacy organization and asks each participant to refer additional participants. Critics say the study is 'not generalizable.' Which response best defends the sampling choice?
AThe study is generalizable because the researcher collected enough interviews to reach saturation
BGeneralizability is the wrong standard here; snowball sampling was necessary because no sampling frame exists for this population, and depth of understanding compensates for breadth
CThe researcher should have combined snowball and stratified sampling to improve representativeness
DSnowball samples become statistically representative once enough referrals are collected
Statistical generalizability requires a probability sample, which presupposes a sampling frame — a list of population members from which to sample randomly. Undocumented immigrants have no accessible sampling frame. Snowball sampling is the appropriate strategy for exactly this situation (hidden or hard-to-reach populations), and it is evaluated by different standards: theoretical depth, internal validity, and whether insights illuminate the phenomenon being studied. Criticizing it for failing probability-sample standards applies the wrong criterion to the wrong method. The researcher's obligation is to justify the strategy in terms of the question, not to claim statistical representativeness.
Question 3 True / False
A large probability survey with 50,000 respondents can produce less valid knowledge about a population than a smaller, carefully executed qualitative study.
TTrue
FFalse
Answer: True
Sample size determines the precision of population estimates only when the measurement itself is valid. If survey questions are poorly worded, leading, or measuring the wrong construct, a large probability sample will precisely estimate the wrong thing. Validity — whether you are measuring what you intend — is independent of sample size. A small qualitative study with rigorous attention to measurement can generate deep, valid insight into phenomena that surveys cannot adequately capture. Large samples reduce sampling error; they do not fix measurement error or conceptual misspecification.
Question 4 True / False
Stratified random sampling's primary purpose is to ensure the sample reflects the demographic diversity of the population.
TTrue
FFalse
Answer: False
Stratified sampling's primary purpose is to improve statistical precision and ensure adequate representation of subgroups for analytical purposes — not to reflect population diversity for its own sake. By sampling within strata (subgroups), researchers reduce variance compared to simple random sampling when the variable of interest differs across strata, and they ensure rare groups appear in sufficient numbers to analyze separately. Simple random sampling already reflects population composition proportionally. Stratification additionally optimizes for precision and analytical control — the goal is statistical efficiency and subgroup analysis, not diversity representation.
Question 5 Short Answer
Why might a qualitative researcher deliberately seek out cases that seem to disconfirm their emerging theoretical framework, rather than focusing on cases that confirm it?
Think about your answer, then reveal below.
Model answer: Seeking disconfirming cases — called negative case analysis — stress-tests and refines the theory. If the theory only explains confirming cases, it may be incomplete or only applicable to a restricted range. A disconfirming case either reveals a flaw requiring revision, exposes an important boundary condition, or shows the exception can be explained by a more nuanced version of the theory. This process mirrors the logic of scientific falsification: theories that survive attempts at disconfirmation are stronger and more credible than theories built only on confirming evidence.
In grounded theory, theoretical sampling explicitly directs the researcher to next cases based on what would most challenge or extend the current conceptual framework. Saturation means new cases add no new conceptual variation — not that every case confirms the theory. Deliberately seeking disconfirming cases accelerates saturation and makes final theoretical claims more defensible and nuanced. It also prevents the common bias of only noticing evidence that fits one's current framework.