On Hypothesis Testing, Confidence Interval, Effect Size and Noncentral Probability Distributions

Abstract

The research provides evidence that null hypothesis significance testing (NHST) is frequently misunderstood and, consequently, misused by researchers. Sources of these misunderstandings can be stated as: 1) Interpreting NHST as a probability on accuracy of null hypothesis, 2) regarding NHST as an indicator of effect size, and 3) overlooking the fact that NHST is just an analysis on means. This study investigates these misuses and discusses confidence interval, effect size and noncentral probability distributions through their possible contributions on resolving the misuses.