A research design is the strategy for a study and the plan by which the strategy is to be carried out. It specifies the methods and procedures for the collection, measurement, and analysis of data. Some major descriptors of designs are: • Exploratory versus formalized. • Observational versus interrogation/communication. • Experimental versus ex post facto. • Descriptive versus causal. • Cross-sectional versus longitudinal. • Case versus statistical. • Field versus laboratory versus simulation. • Subjects' perceptions: no deviations, some deviations, or researcher-induced deviations.
Exploratory research is appropriate for the total study in topic areas where the developed data are limited. In most other studies, exploration is the first stage of a project and is used to orient the researcher and the study. The objective of exploration is the development of hypotheses, not testing. Formalized studies, including descriptive and causal, are those with substantial structure, specific hypotheses to be tested, or research questions to be answered. Descriptive studies are those used to describe phenomena associated with a subject population or to estimate proportions of the population that have certain characteristics. Causal studies seek to discover the effect that a variable(s) has on another (or others) or why certain outcomes are obtained.
The relationships that occur between two variables may be symmetrical, reciprocal, or asymmetrical. Of greatest interest to the research analyst are asymmetrical relationships, which may be classified as any of the following types: • Stimulus-response • Property-disposition • Disposition-behavior • Property-behavior We test causal hypotheses by (1) measuring the covariation among variables, (2) determining the time order relationships among variables, and (3) ensuring that other factors do not confound the explanatory relationships. Where possible, we try to achieve the ideal of experimental design with random assignment of subjects, matching of subject characteristics, and manipulation and control of variables. Using these methods and techniques, we measure relationships as accurately and objectively as possible.