Evaluating the statistical power of goodness-of-fit tests for health and medicine survey data

Abstract

Goodness-of-fit test statistics are widely used in health and medicine related surveys however little regard is usually given to their statistical power. This paper investigates the simulated power of five categorical goodness-of-fit test statistics used to analyze health and medicine survey data collected on a 5-point Likert scale. The test statistics used in this power study are Pearson's Chi-Square, the Kolmogorov-Smirnov test statistic for discrete data, the Log-Likelihood Ratio, the Freeman-Tukey and the special case of the Power Divergence statistic defined by Cressie and Read (1984). Recommendations based on these simulations are provided on which of these goodness-of-fit test statistics is the most powerful overall and which is the most powerful for the predefined uniform null against the four general shaped alternative distributions (see Figure 1) investigated in this paper.