Work through the questions before checking solutions. Guide question labels stay the same when you shuffle the bank. Use an exact fraction when a decimal would require rounding.
Guide question labels stay the same when the order changes. Check answers when you are ready to review.
Written responses are self-review exercises.
Saved on this browser and device; clearing browser data removes progress.
The drill
Your practice is saved on this device.
Question 1
Guide question E7A
Researchers randomly sample adults from a town, record self-selected reading habits and vocabulary scores, and find an association. All selected adults respond. Which inference is supported by this design?
Why this answer
Answer: A
Random sampling supports generalizing the observed association to the town’s adults. Reading habits were observed, not assigned, so other factors may explain some or all of the association. Random assignment is not required to describe an association, but is central to the randomized causal comparisons taught here.
Volunteers are randomly assigned to two training programs. The study is sound and finds convincing evidence of a mean improvement difference. Which conclusion is best?
Why this answer
Answer: D
Random treatment assignment can support a causal comparison among the participating volunteers. It does not make their recruitment representative of all people. Separate the assignment stage from the sampling stage.
Which feature most directly reduces systematic confounding in a treatment comparison?
Why this answer
Answer: B
Random assignment tends to balance other characteristics across treatment groups, preventing treatment from being systematically assigned to one kind of participant. It does not guarantee exact balance or a successful treatment. Selecting only successful participants would introduce bias.
Only 30 of 200 randomly selected people respond to a survey. Which concern is most relevant?
Why this answer
Answer: C
The initial random sample can be legitimate while the final responding subset is systematically different. This creates possible nonresponse bias. Neither its existence nor its direction is established merely by the low response count, but it is an important concern.
An instructor uses a new method only in morning classes and the old method only in afternoon classes. The morning classes perform better. Which limitation is most direct?
Why this answer
Answer: A
The treatment groups differ systematically in both method and class time; student composition may also differ. The observed advantage cannot be assigned solely to the method from this comparison. A better design would randomize methods within relevant class-time groups when feasible.