Location Tests for Biomarker Studies: A Comparison Using Simulations for the Two-sample Case^[*]

Summary

Background: Gene, protein, or metabolite expression levels are often non-normally distributed, heavy tailed and contain outliers. Standard statistical approaches may fail as location tests in this situation.

Objectives: In three Monte-Carlo simulation studies, we aimed at comparing the type I error levels and empirical power of standard location tests and three adaptive tests [O’Gorman, Can J Stat 1997; 25: 269 –279; Keselman et al., Brit J Math Stat Psychol 2007; 60: 267– 293; Szymczak et al., Stat Med 2013; 32: 524 – 537] for a wide range of distributions.

Methods: We simulated two-sample scena -rios using the g-and-k-distribution family to systematically vary tail length and skewness with identical and varying variability between groups.

Results: All tests kept the type I error level when groups did not vary in their variability. The standard non-parametric U-test per -formed well in all simulated scenarios. It was outperformed by the two non-parametric adaptive methods in case of heavy tails or large skewness. Most tests did not keep the type I error level for skewed data in the case of heterogeneous variances.

Conclusions: The standard U-test was a powerful and robust location test for most of the simulated scenarios except for very heavy tailed or heavy skewed data, and it is thus to be recommended except for these cases. The non-parametric adaptive tests were powerful for both normal and non-normal distributions under sample variance homogeneity. But when sample variances differed, they did not keep the type I error level. The parametric adaptive test lacks power for skewed and heavy tailed distributions.

Keywords

Adaptive location tests - adaptive trimming - g-and-k-distribution family - linear rank tests - selector statistics

^* Supplementary material published on our website www.methods-online.com

• References

• 1 Szymczak S, Scheinhardt MO, Zeller T, Wild PS, Blankenberg S, Ziegler A. Adaptive linear rank tests for eQTL studies. Stat Med 2013; 32: 524-537.
• 2 Fagerland MW, Sandvik L. The Wilcoxon-Mann-Whitney test under scrutiny. Stat Med 2009; 28: 1487-1497.
• 3 Szymczak S, Igl B-W. Ziegler A. Detecting SNP-expression associations: A comparison of mutual information and median test with standard statistical approaches. Stat Med 2009; 28: 3581-3596.
• 4 O’Gorman TW. An adaptive test for the one-way layout. Can J Stat 1997; 25: 269-279.
• 5 Keselman HJ, Wilcox RR, Lixl LM, Algina J, Fradettel K. Adaptive robust estimation and testing. Brit J Math Stat Psychol 2007; 60: 267-293.
• 6 Haynes MA, MacGillivray HL, Mengersen KL. Robustness of ranking and selection rules using generalised g-and-k distributions. J Plan Stat Inf 1997; 65: 45-66.
• 7 Hogg RV. Adaptive robust procedures: a partial review and some suggestions for future applications and theory. J Am Stat Assoc 1974; 69: 909-923.
• 8 Reed JF, Stark DB. Hinge estimators of location: Robust to asymmetry. Comput Methods Programs Biomed 1996; 49: 11-17.
• 9 Büning H. Adaptive tests for the c-sample location problem - the case of two-sided alternatives. Commun Stat Theor Meth 1996; 25: 1569-1582.
• 10 Beier F. Adaptive Tests bei nicht-monotonen Dosis-Wirkungsbeziehungen. Dortmund: Technische Universität; 1996.
• 11 Handl A. Masszahlen zur Klassifizierung von Verteilungen bei der Konstruktion adaptiver verteilungsfreier Tests im unverbundenen Zweistichproben-Problem. Berlin: Freie Universität Berlin; 1985.
• 12 Bradley JV. Robustness?. Br J Math Stat Psychol 1978; 31: 144-152.