ERIC Number: ED201658
Record Type: RIE
Publication Date: 1981-Apr-14
Reference Count: 0
A Comparison of Six Robust Correlation Estimators.
Baranowski, B. Bonnie; Halperin, Silas
A Monte Carlo investigation of six robust correlation estimators was conducted for data from distributions with longer than Gaussian tails: a bisquare coefficient, the Tukey correlation, the standardized sums and differences, a biweight standardized sums and differences, the transformed Spearman's rho and a bivariate trimmed Pearson. Evaluation of the estimators was based on bias and variability as measured by mean square error, and efficiency relative to the Pearson correlation coefficient. Correlations of .9, .3, .6, and .0 for samples of size twenty and forty for different weight tails were used. Two estimators that stood out as not being situation specific and demonstrated robust properties were the averaged bisquare, and Tukey correlations. The transformed Spearman's rho was easy to calculate, and for small correlations was clearly the best of all the estimators considered; however, it degenerated as the correlation increased so that it was among the worst for a correlation of .9. The standardized sums and differences estimator demonstrated robustness and warrants further investigation. The biweight standardized sums and differences estimator did not demonstrate robustness, and the bivariate trimming estimator demonstrated limited usefulness. (Author/RL)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: Estimation (Mathematics); Monte Carlo Studies; Robustness; Variability
Note: Paper presented at the Annual Meeting of the American Educational Research Association (65th, Los Angeles, CA, April 13-17, 1981).