Sampling Flashcards
What’s the difference between probability and non-probability sampling?
Former uses a statistical method to select a sample from the population; the former is not statistical.
What is purposive sampling? Benefits and drawbacks?
The non-statistical selection of members of the population who have been selected deliberately.
Benefit is taking advantage of some known variation or sample characteristic. May represent a “best case” situation. inexpensive.
Drawbacks: experimental artificiality, selection bias (pre-existing differences), effect of prior treatment (if applicable), unrepresentativeness/ sampling bias
What is convenience sampling? Benefits and drawbacks?
Selecting participants owing to the ease of accessibility.
Benefit is taking advantage of some known variation or sample characteristic. May represent a “best case” situation.
Drawbacks: experimental artificiality, selection bias (pre-existing differences), effect of prior treatment (if applicable), unrepresentativeness/ sampling bias
What is volunteer sampling? Benefits and drawbacks?
Selection of participants based upon those who volunteer to be part of the study.
Benefits: may be more politically feasible, cooperative respondents
Drawbacks: Major selection problems, unrepresentativeness/ sampling bias
What is haphazard sampling? Benefits and drawbacks?
A sampling of available participants. A subset of volunteers.
Benefits: political feasibility, possibly exploit some known variation
Drawbacks: selection issues, unrepresentativeness/ sampling bias
What is random sampling? Benefits and drawbacks?
The selection of participants at random from the population.
on expectation should not be any internal threats.
may be drawn from a limited portion of pop, unrepresentative. expensive.
What is stratified sampling? When is it needed?
Categorizing members of the population into groups and selecting from those groups. Needed if the population is heterogeneous wrt variables of interest.
benefits: guarantees adequate representation of interesting subgroups. reduces error in estimations of these subgroups
drawback: only equates the groups on the stratified variable, other differences among groups are not equated.
What is cluster sampling?
Selection of participants from within established groups. Sampling blocks, then sampling a house from within the block; sampling schools, then sampling teachers/ students within the schools.
less expensive
less precise SE
What is quota sampling? Benefits and drawbacks?
A stratification of the population, that’s then sampled using convenience.
more likely to be willing
unrepresentative
What is systematic sampling?
Selecting participants at given intervals (e.g. every 4th person)
convenient, inexpensive
possible periodicity in list (non-random)
What is multistage area sampling?
A variant of cluster sampling, nested. Grades within schools within districts.
less expensive than random sampling, more power than clustering. equal in expectation across clusters, then within clusters. ideally, individuals could be randomly assigned to clusters, then clusters randomly assigned to treatment.
higher error than random sampling, higher cost than clustering
What are some common sampling mistakes?
R
Failure to define the accessible and target populations, and provide evidence of their similarities;
sample is too small for stats analysis;
incomplete understanding of differences between volunteer and non-volunteers;
failure to account for attrition;
non-equivalent groups
failure to examine attrition
no examination of representativeness
What is “functionally random assignment” and why does it seem innocuous, but is actually a bad idea?
Functionally random assignment may seem random (e.g. assigning the odd numbered SSN to treatment and evens to control) but is not, so there may be systematic selection bias.
What is the problem with an N=1 for each treatment status? What if one cluster level administers both treatments?
Cluster level confounders cannot be controlled for if treatment is administered at a cluster level, even if students are randomly assigned to each cluster.
The cluster level agents (schools, teachers, etc.) could not vary in their motivation, ability, etc. regarding the administration of treatment among groups, otherwise the cluster is confounded.
when writing a proposal what aspects about conducting research at a site should be addressed?
location, cooperation with local leaders, dates, recruitment/ sampling, collection procedure
