Measure of Location: Comparing Means, Trimmed Means, One Step M-estimators and Modified One Step M-estimators under Non-normality

集中量數：平均數、削截平均數、單階M估計值和改良式單階M估計值之比較

10.6129/CJP.2004.4601.03

吳佩真(Pei-Chen Wu)

集中量數 ； 平均數 ； 削截平均數 ； one step M-estimators based ； modified one step M-estimators ； measure of location ； means ； trimmed means ； one step M-estimators ； modified one step M-estimators

中華心理學刊

46卷1期（2004 / 03 / 01）

29 - 47

英文

本研究比較五種集中量數（平均數、10%和20%削截平均數、one step M-estimators based on Huber's Ψ、modified on step M-estimators）如何受非常態分佈的影響。由24組實際資料之檢驗結果發現：當比較二組獨立團體的集中量數時，20%削截平均數、one step M-estimator與modified one step M-estimator是較佳的選擇，而平均數是最差之選擇，因為平均數極容易受某一極端值的影響。此外，研究亦發現，集中量數僅是反應一個群體中心位置的數值。因此，若是僅以集中量數來比較二組獨立群體，將不易察覺到二組群體以複雜方式呈現不同，故本研究建議比較二組的quantiles才可完整、全盤得知二組群體間的關係。

This study examines five measures of location (means, 10% and 20% trimmed means, one step M-estimators based on Huber's Ψ and modified one step M-estimators) in terms of their Type I error rates, standard errors and significance levels in 24 empirical data sets. Twenty-four empirical data sets can be categorized into eight kinds of distributions which frequently arise in educational and psychological research- discrete mass at zero, mass at zero, extreme positive skew, extreme negative skew, bimodality, multi-modality and lumpy, digit preference, and smooth symmetric. The results show that the 20% trimmed mean, one step M-estimator and modified one step M-estimator, are good alternatives for comparing two groups based on comparing measure of location. Student's t is the least satisfactory statistic. Also, this study indicates that comparing measures of location provides information only on the typical value of the groups. This limitation is apparent in some situations considered here where none of the five measures of locations are completely satisfactory. Thus, the study recommends comparing quantiles of two groups to obtain an overall picture of the relationship between two groups.

1. Bradley, J. V.(1978).Robustness?.British Journal of Mathematical and Statistical Psychology,31,144-152.
2. Doksum, K. A.(1974).Empirical probability plots and statistical inference for non-linear models in the two-sample case.Annals of Statistics,2,267-277.
3. Emerson, J. D.,Stoto, M., A.(2000).Understanding robust and exploratory data analysis.New York:John Wiley & Sons, Inc..
4. Geary, R. C.(1947).Testing for normality.Biometrika,34,209-242.
5. Hall, P.,Padmanabhan, A. R.(1992).On the bootstrap and the trimmed mean.Journal of Multivariate Analysis,41,132-153.
6. Huber, P. J.(1981).Robust statistics.New York:Wiley.
7. Huber, P. J.(1964).Robust estimation of location.Annals of Mathematical Statistics,35,73-101.
8. Huber, P. J.(1993).New directions in statistical data analysis and robustness.Boston:Birkhauser Verlag.
9. Lix, A. M.,Keselman H. J.(1998).To trim or not to trim: Tests of location equality under heteroscedasticity and nonnormality.Educational and Psychological Measurement,58(3),409-429.
10. Micceri, T.(1989).The unicorn, the normal curve and other improbable creatures.Psychological Bulletin,105(1),156-166.
11. Pearson, E. S.,Please, N. W.(1975).Relation between the shape of population distribution and the robustness of four simple test statistics.Biometrika,62,223-241.
12. Rasmussen, J. L.(1989).Data transformation, type I error rate and power.British Journal of Mathematical and Statistical Psychology,42,203-211.
13. Rosenberger, J. L.,Gasko, M.(2000).Understanding robust and exploratory data analysis.New York:John Wiley & Sons, Inc..
14. Stigler, S. M.(1977).Do robust estimators work with real data?.The Annals of Statistics,5(6),1055-1098.
15. Tukey, J. W.(1970).Exploratory data analysis.
16. Walberg, H. J.,Strykowski, B. F.,Rovai, E.,Hung, S. S.(1984).Exceptional performance.Review of Education Research,54(1),87-112.
17. Wilcox, R. R.(1992).Comparing one-step M-estimators of location corresponding to two independent groups.Psychometrika,57(1),141-154.
18. Wilcox, R. R.(1990).Comparing the means of two independent groups.Biometrical Journal,32,771-780.
19. Wilcox, R. R.(1997).Introduction to robust estimation and hypothesis testing.San Diego, CA:Academic Press.
20. Wilcox, R. R.(1998).The goals and strategies of robust methods.Journal of Mathematical and Statistical Psychology,51,1-39.
21. Wilcox, R. R.(2001).Fundamentals of modern statistical methods: Substantially improving power and accuracy.New York:Springer-Verlag.
22. Wilcox, R. R.(1993).Comparing one-step M-estimators of location when there are more than two groups.Psychometrika,58(1),71-78.
23. Wilcox, R. R.(1998).How many discoveries have been lost by ignoring modern statistical methods?.American Psychologist,53(3),300-314.
24. Wilcox, R. R.(1994).A one way random effects model for trimmed means.Psychometrika,59(3),289-306.
25. Wilcox, R. R.(1992).Why can methods for comparing means have relatively low power, and what can you do to correct the problem?.Current Directions in Psychological Science,1,101-105.
26. Wilcox, R. R.(1987).New statistical procedures for the social sciences: Modern solutions to basic problems.Hillsdale, New Jersey:Lawrence Erlbaum.
27. Wilcox, R. R.(2003).Applied contemporary statistical methods.New York:Academic Press.
28. Wilcox, R. R.(1994).Some results on the Tukey-McLaughlin and Yuen methods for trimmed means when distributions are skewed.Biometrical Journal,36,259-273.
29. Wilcox, R. R.(1995).Simulation results on solutions to the multivariate Behrens-Fisher problem via trimmed means.Statistician,44(2),213-225.
30. Wilcox, R. R.(1996).Statistics for the social sciences.San Diego, CA:Academic Press.
31. Yuen, K. K.(1974).The two-sample trimmed t for unequal population variances.Biometrika,61,165-170.