A class of naive estimators of correlation was tested for robustness, accuracy, and efficiency against Pearson's r, Tukey's rt, and Spearman's ro. It was found that this class of estimators seems in some respects to be superior being less affected by outliers, reasonably efficient, and frequently more easily calculated. The definition and details of the use of these naive estimators are the subject of this paper.
Get full access to this article
View all access options for this article.
References
1.
Box, G.E.P., & Tiao, G.C.A Bayesian approach to some outlier problemsBiometrika , 1968, 55, 119-129.
Devlin, S.J., Gnanadesikan, R., & Kettenring, J.R.Robust estimation and outlier detection with correlation coefficientsBiometrika, 1975, 62, 531-546.
4.
Dixon, W.J.Simplified estimation from censored normal samplesAnnals of Mathematical Statistics, 1960, 31, 385-391.
5.
Downton, F.Linear estimates with polynomial coefficientsBiometrika , 1966, 53, 129-141.
6.
Einhorn, H.J., & Hogarth, R.M.Unit weighting schemes for decision makingOrganizational Behavior and Human Performance, 1975, 13, 171-192.
7.
Hogg, R.Adaptive robust procedures: A partial review and some suggestions for further applications and theoryJournal of the American Statistical Association, 1974, 69, 909-927.
8.
Kahneman, D., & Tversky, A.On the psychology of predictionPsychological Review , 1973, 80, 237-251.
9.
Meehl, P.E.Clinical versus statistical prediction: A theoretical analysis and a review of the evidenceMinneapolis: University of Minnesota Press, 1954.
Tukey, J.W., & McLaughlin, D.H.Less vulnerable confidence and significance procedures for location based on a single sample: Trimming/Winsorization 1Sankhya, 1963, A25, 331-352.
12.
Wainer, H.Estimating coefficients in linear models: It don't make no nevermindPsychological Bulletin, 1976 , 83, 213-217.
