NONPARAMETRIC INDEXES FOR SENSITIVITY AND BIAS: COMPUTING FORMULAS

NONPARAMETRIC INDEXES FOR SENSITIVITY AND BIAS: COMPUTING FORMULAS

1971, Vol. 73, No. 6, 424-429 | J. BROWN GRIER
The article by J. Brown Grier from Northern Illinois University presents computing formulas for two nonparametric indexes of sensitivity and bias used in signal detectability studies. The sensitivity index is related to \(P(I)\), a statistic with known sampling variability. A new bias index is proposed, which overcomes certain inconveniences while yielding identical isobias contours. The formulas for these indexes are derived, and their relationship to the area measure under the operating characteristic curve is established. The article also provides explicit expressions for the isosensitivity and isobias contours. Examples using data from Green and Swets (1966) and Murdock (1965) illustrate the application of these indexes. The new indexes are shown to be informative and easily computable, making them useful for analyzing data without specific assumptions about underlying distributions. The discussion highlights the practical utility of these indexes and the need for further statistical analysis to combine multiple data points into a common area estimate.The article by J. Brown Grier from Northern Illinois University presents computing formulas for two nonparametric indexes of sensitivity and bias used in signal detectability studies. The sensitivity index is related to \(P(I)\), a statistic with known sampling variability. A new bias index is proposed, which overcomes certain inconveniences while yielding identical isobias contours. The formulas for these indexes are derived, and their relationship to the area measure under the operating characteristic curve is established. The article also provides explicit expressions for the isosensitivity and isobias contours. Examples using data from Green and Swets (1966) and Murdock (1965) illustrate the application of these indexes. The new indexes are shown to be informative and easily computable, making them useful for analyzing data without specific assumptions about underlying distributions. The discussion highlights the practical utility of these indexes and the need for further statistical analysis to combine multiple data points into a common area estimate.
Reach us at info@study.space