Research methods in any field of inquiry can involve two types of reasoning: inductive and deductive. While inductive reasoning employs a more exploratory, openended approach, working from specific examples toward development of a theory, deduction is more narrow and focused on testing specific hypotheses. Some scientists characterize deductive reasoning as a topdown approach to research, in which researchers proceed from the general to the specific.
Broad Theory

A deductive approach to research begins with a general theory or question about a subject of interest. For example, an economist might be interested in the extent to which education influences an individual's earnings. The theory may hold that the more education a person acquires, the more money she earns.
Hypothesis Generation

A deductive approach requires a researcher to proceed from the broad theory to the development of a testable hypothesis. In the education and earnings example, an economist could develop a hypothesis about the relationship between education, measured as years of schooling, and earnings, measured as a person's annual salary.
Data Collection

After developing a hypothesis that includes measurable variables, a researcher then collects the data necessary to test that hypothesis. In the education and earnings example, an economist might collect data on individuals' earnings and years of education from a reputable data source, such as the federal government's Bureau of Labor Statistics. The data in this example would allow a researcher to employ the statistical and quantitative research methods often associated with the deductive method to test the hypothesis.
Hypothesis Testing and Conclusions

After data collection, deductive research proceeds to hypothesis testing. In research studies that use statistical or quantitative methods, researchers often state their beginning hypothesis (known in statistics as a null hypothesis) in the negative. In this example, the null hypothesis would be that years of schooling has no effect on a person's earnings. If the hypothesis test reveals an impact, whether positive or negative, the researcher rejects the null hypothesis of no effect, provided the observed effect is statistically significant. A statistically significant effect is one in which the researcher is at least 95 percent certain that the relationship between two variables (education and earnings in this example) is not due to random chance. It is important to note that statistical significance does not prove causation. The results of hypothesis testing under the deductive approach often lead researchers to generate new theories.
References
 Deductive and Inductive Research Approach
 The Practice of Social Research (8th ed.), Earl Babbie, 1998