A research design is a plan of action for executing a research project that specifies the theory to be tested, the unit of analysis (such as individual, organization, or country), the necessary observable data and how it will be collected, and the procedures that will be used to examine the data.
- All the parts of a research design should work to the same end: drawing sound conclusions supported by observable evidence.
- Many things affect the choice of research design including the purpose of the research (exploratory, descriptive, explanatory) and practical limitations (resources such as time, money, and skill or ethical concerns).
Correlation means that two things are related. Causation means that a change in the state of one thing brings about (in full or in part) a change in the state of another.
Causal relationships meet three characteristics:
- Covariation: the alleged cause varies with the supposed effect.
- Time order: the cause precedes the effect in time.
- Elimination of alternative explanations: the research design should eliminate as many alternative explanations for the observed effect as possible to isolate causation to one factor.
The classical randomized experiment has five basic characteristics. The researcher:
- Establishes at least one experimental group that will have exposure to the treatment and one control group that will not.
- Randomly assigns individuals to each group, avoiding self-selection, and guaranteeing that on average the groups will not differ in any respect.
- Controls the administration of the treatment including the circumstances, under which the experimental group is exposed.
- Establishes and measures a dependent variable before and after the treatment with a pre-test and post-test. Because treatment groups are constituted through randomization, any difference between the pre- and post-tests can be attributed to the experimental effect of exposure to the treatment.
- Controls the environment of the experiment (time, location, and other physical aspects).
Two important factors in judging the quality of a research design with respect to causal relationships are internal and external validity.
- Internal validity refers to a causal relationship that was not created by a spurious relationship.
- External validity refers to the extent to which the results of an experiment can be generalized across populations, time, and settings.
Other randomized experiments include the post-test design, the repeated-measurement design, the multiple group design, and field experiments, are variations on the classical randomized experiment, and may be more practical for some situations. Nonrandomized designs, such as natural experiments, are quasi-experiments.
Intervention analyses use measurements of a dependent variable both before and after the introduction of an independent variable that is observed but not controlled by the researcher.
Observational study describes quasi-experimental designs in which the researcher does not manipulate experimental variables or randomly assign subjects to treatments but instead merely observes causal sequences and covariations.
Examples of observational study designs in this chapter include:
- Small-N designs; deep understanding of a small number of cases.
- Comparative designs that compare a small number of cases in detail.
- Focus groups: used to create hypotheses for testing through other research designs.
- Cross-sectional designs; characterized by measurements of the independent and dependent variables at approximately the same time including surveys and aggregate data analysis.
- Longitudinal designs; allow for the measurement of variables at different points in time including trend analysis and panel studies.
- While experimental research is generally very good with respect to internal validity and suffers more problems with external validity, nonexperimental designs with the exception of small-N designs are generally characterized as having less internal reliability but better external validity.