HLT 3672 Topic 4 DQ 1: Research Design Types

The three designs are separated by two questions, and stating them first turns a list into a framework. Does the researcher control who receives the intervention, and is assignment to condition random? Experimental designs answer yes to both, which is what supports a causal claim, because randomisation distributes both known and unknown confounders across groups. Quasi-experimental designs manipulate the intervention but assign by something other than chance — an existing group, a clinic, a time period — so they support a weaker causal claim and require the author to argue that the groups were comparable. Nonexperimental designs observe without manipulating anything, and describe or correlate rather than intervene. Use those two questions to classify each example you find, and say which answer each study gives. Say which answer each of your three studies gives to both questions, in one line each, before you discuss any of them in detail. That table is the comparison the prompt is asking for.

Then find real studies, since the prompt asks for examples from the library and a textbook description of a design does not satisfy it. Read the methods section rather than the abstract, because papers describe themselves loosely and a study calling itself a trial may have assigned by ward. For each, say what the design allows the authors to claim and what it does not: a randomised trial can support a causal statement in the studied population, a quasi-experiment can support one only if the comparison group is defensible, and a cross-sectional study cannot establish temporal order at all, so it cannot distinguish cause from consequence. Close on the trade-off, which is the mature point: the stronger the design for causal inference, the narrower and more artificial the sample tends to be, so the design that best answers whether an intervention works may be the worst at telling you whether it will work in your population.