empirical evidence

empirical evidence, information gathered directly or indirectly through observation or experimentation that may be used to confirm or disconfirm a scientific theory or to help justify, or establish as reasonable, a person’s belief in a given proposition. A belief may be said to be justified if there is sufficient evidence to make holding the belief reasonable.

The concept of evidence is the basis of philosophical evidentialism, an epistemological thesis according to which a person is justified in believing a given proposition p if and only if the person’s evidence for p is proper or sufficient. In this context, the Scottish Enlightenment philosopher David Hume (1711–76) famously asserted that the “wise man…proportions his belief to the evidence.” In a similar vein, the American astronomer Carl Sagan popularized the statement, “Extraordinary claims require extraordinary evidence.”

Foundationalists, however, defend the view that certain basic, or foundational, beliefs are either inherently justified or justified by something other than another belief (e.g., a sensation or perception) and that all other beliefs may be justified only if they are directly or indirectly supported by at least one foundational belief (that is, only if they are either supported by at least one foundational belief or supported by other beliefs that are themselves supported by at least one foundational belief). The most influential foundationalist of the modern period was the French philosopher and mathematician René Descartes (1596–1650), who attempted to establish a foundation for justified beliefs regarding an external world in his intuition that, for as long as he is thinking, he exists (“I think, therefore I am”; see cogito, ergo sum). A traditional argument in favour of foundationalism asserts that no other account of inferential justification—the act of justifying a given belief by inferring it from another belief that itself is justified—is possible. Thus, assume that one belief, Belief 1, is justified by another belief, Belief 2. How is Belief 2 justified? It cannot be justified by Belief 1, because the inference from Belief 2 to Belief 1 would then be circular and invalid. It cannot be justified by a third nonfoundational Belief 3, because the same question would then apply to that belief, leading to an infinite regress. And one cannot simply assume that Belief 2 is not justified, for then Belief 1 would not be justified through the inference from Belief 2. Accordingly, there must be some beliefs whose justification does not depend on other beliefs, and those justified beliefs must function as a foundation for the inferential justification of other beliefs.

Empirical evidence can be quantitative or qualitative. Typically, numerical quantitative evidence can be represented visually by means of diagrams, graphs, or charts, reflecting the use of statistical or mathematical data and the researcher’s neutral noninteractive role. It can be obtained by methods such as experiments, surveys, correlational research (to study the relationship between variables), cross-sectional research (to compare different groups), causal-comparative research (to explore cause-effect relationships), and longitudinal studies (to test a subject during a given time period).

Qualitative evidence, on the other hand, can foster a deeper understanding of behaviour and related factors and is not typically expressed by using numbers. Often subjective and resulting from interaction between the researcher and participants, it can stem from the use of methods such as interviews (based on verbal interaction), observation (informing ethnographic research design), textual analysis (involving the description and interpretation of texts), focus groups (planned group discussions), and case studies (in-depth analyses of individuals or groups).

Empirical evidence is subject to assessments of its validity. Validity can be internal, involving the soundness of an experiment’s design and execution and the accuracy of subsequent data analysis, or external, involving generalizability to other research contexts (see ecological validity).