Strong quasi-experimental designs for research on the effectiveness of rehabilitation.

Medical rehabilitation needs better understanding of the effectiveness of its treatments and of patient characteristics most responsive to alternative intervention strategies. The goal of this paper is to improve understanding of research design in medical rehabilitation. More specifically, it describes two potentially rigorous but infrequently used "quasi-experimental" research designs--the regression-discontinuity design and the multiple interrupted time-series design. These are contrasted with the strongest research design--the randomized experiment--and to weaker designs, such as the nonequivalent group designs. Pre-experimental research, including qualitative, descriptive, and predictive studies, should not be confused with experimental research designs. More frequent use of randomized experimental and strong quasi-experimental designs can provide knowledge that will augment the effectiveness of rehabilitation practice.