Randomized controlled trials (RCTs) are said to be the gold-standard method in evaluation research. What if I cannot perform the random assignment in my sampling of street sleepers?

Glad to hear your feedback.

That makes sense. Great!

Indeed, it is more convincing if RCTs are feasible. However, in practice, we often are unable to conduct an RCT study, limited by the resources and other situations. So then, a quasi-experimental design with a nonequivalent comparison group is also recommendable.