Scientific Method

Inductive and Deductive Methods

Inductive reasoning uses specific instances to infer general principles, whereas deductive reasoning derives specific conclusions from one or more premises or axioms. Inductive reasoning takes many forms, but a popular inductive approach [Page 1473]comes up with a generalization based on a set of observations. An historical example of this method is the investigation by physician John Snow of the 1854 outbreak of cholera in London, which killed hundreds in a few weeks. Many explanations were offered for this outbreak. No one really understood how diseases spread, as the germ theory of disease had yet to be proposed (that happened after 1860 with Louis Pasteur’s study of puerperal fever). The prevailing view was something called the miasma theory, which held that “noxious exhalations” from swamps and like places somehow cause disease.

In those days, there was no domestic water supply. People got their water from hand pumps scattered across the city. The pumps had different sources—local wells or piped from the river Thames or one of its tributaries. Snow did not believe in the miasma theory and sought another explanation for the spread of the disease. He looked at where cases of cholera had occurred. He noticed that almost all of them were clustered within walking distance of a particular hand pump in Broad Street. This allowed him to come up with a hypothesis, that the water from the Broad Street pump is contaminated in some way that causes cholera. The hypothesis suggested an obvious experimental test: Remove the handle from the Broad Street pump so that no water can be obtained from it. Snow managed to persuade the local council to disable the pump. His hypothesis was confirmed: The incidence of cholera dropped dramatically, proving (without the aid of statistics) that the pump was the source of the disease.

Snow’s experiment is what is called an AB design: Two conditions/treatments are applied in succession—handle/no-handle. In a laboratory context, both conditions would normally be repeated, ABAB, just to be sure that B really has the predicted effect. In Snow’s case, this was both unnecessary (the effect of removing the handle was large) and unethical (restoring the handle might have caused more deaths)—and in any case, his main purpose was to improve public health rather than advance knowledge.

...