Research Design
State the logic that ensures the data validly answers the research question
The research design is the method logic by which the data collected answers the research question. The concept applies to exploratory and correlational research, but is mostly used in explanatory research, which involves an intervention, i. e. we do something, then see what happens. While "correlation does not imply causality, as if we just measure two things we don't know which came first, in explanatory research we know what came first because we did it. So any correlation results are either causality or chance. The various explanatory research designs are all trying to answer the question "Was it just coincidence?":
Experimental designs require that subjects be randomly assigned to "interventions" or "treatments" (what you do), e. g. so you cant let subjects evaluating web sites choose which site they evaluate, lest any differences found are due to the spurious cause of their preference. If two groups of subjects rating two web sites, one assumes they are "matched", lest any differences found be due to differences in the groups not the web sites.
Quasi-experimental designs are methods where subjects cannot be assigned randomly, e. g. the effect of male-female gender.
Repeated measures designs subjects try first one treatment then another, e. g. try one software product then another. Here the two trials involve the same subjects, who obviously match themselves. However there can be order effects, so the order subjects try the software must be randomized to avoid it affecting the results, e.g. subjects may be more tired on the second test.
Stating the research design logic may only take a sentence or two, but is very important. Some explanatory designs are:
One-shot case study: Or ex post facto design, where a single group of subjects is measured after some intervention, e. g. studying the effect of new security measures. However even if subjects liked the new measures, perhaps they equally liked the old ones?
Two-group post-test only without random assignment: In this static (one-time) design, a group without the intervention is also measured, e. g. smokers are compared to non-smokers for health effects. The problem here is that people with weak health initially may be precisely those who choose smoke. For example soldiers about to enter battle may smoke, as they figure they wont last long anyway, but iit is not smoking that causes their death.
One-group pretest-posttest: A single group of subjects is measured before and after some intervention, e. g. measure health before and after an exercise program. Here we know the measure subjects began with, but it has no control for research bias or time trends. However if many other things happened at the same time, who is to say it was the security measures that caused any effects found?
Two-group pretest-posttest: In this design subjects are randomly assigned to a treatment group and a control group. Both are measured before and after some intervention, e. g. to assess a new training web site, the treatment group will try it while the control group sticks to the old one. This controls for everything but expectancy bias.
Two-group pretest-posttest without random assignment: Same as #4 but quasi-experimental (non-random assignment).
Solomon four-group design: The most complex design, as it also controls for experimental bias, as two groups don't do the pretest. It takes a lot of work to do.
Example(s)
(Use a descriptive name, e. g. "ITExample". Or click on an existing collection and edit it.)