HomeKAYÂ TAOvol. 12 no. 1 (1993)

Some Burning Issues in Statistical Inference in the Behavioral Sciences

Ben Teehankee

Discipline: Psychology, Social Science



Classical hypothesis testing (also called statistical significance testing) is well-entrenched in behavioral science research. Most journals in psychology and sociology feature articles utilizing one form of statistical test or another. A course in statistical hypotheses testing is de rigueur in undergraduate curricula in the behavioral sciences. Even students writing their theses pin their hopes on the result of their statistical tests.


Amidst this widespread use of statistical significance tests, it has become easy to disregard a number of issues related to such tests. The objective of this paper is to review some of these more salient issues in the interest of putting statistical tests in a more proper perspective given the objectives of the behavioral science research enterprise.