Bias and prevalence effects on kappa viewed in terms of sensitivity and specificity

被引:208
作者
Hoehler, FK [1 ]
机构
[1] Data Management Ctr, Orange, CA 92868 USA
关键词
kappa; sensitivity; specificity; ROC; bias; prevalence;
D O I
10.1016/S0895-4356(99)00174-2
中图分类号
R19 [保健组织与事业(卫生事业管理)];
学科分类号
摘要
Paradoxical effects of bias and prevalence on the kappa coefficient are examined using the concepts of sensitivity and specificity. Results that appear paradoxical when viewed as a 2 x 2 table of frequencies do not appear paradoxical when viewed as a pair of sensitivity and specificity measures where each observer is treated as a predictor of the other observer. An adjusted kappa value can be obtained from these sensitivity/specificity measures but simulation studies indicate that it would result in substantial overestimation of reliability when bias or prevalence effects are observed. It is suggested that investigators concentrate on obtaining populations with trait prevalence near 50% rather than searching fur statistical indices to rescue or excuse inefficient experiments. (C) 2000 Elsevier Science Inc. All rights reserved.
引用
收藏
页码:499 / 503
页数:5
相关论文
共 11 条
[1]   BIAS, PREVALENCE AND KAPPA [J].
BYRT, T ;
BISHOP, J ;
CARLIN, JB .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 1993, 46 (05) :423-429
[2]   HIGH AGREEMENT BUT LOW KAPPA .2. RESOLVING THE PARADOXES [J].
CICCHETTI, DV ;
FEINSTEIN, AR .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 1990, 43 (06) :551-558
[3]   A COEFFICIENT OF AGREEMENT FOR NOMINAL SCALES [J].
COHEN, J .
EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT, 1960, 20 (01) :37-46
[4]   HIGH AGREEMENT BUT LOW KAPPA .1. THE PROBLEMS OF 2 PARADOXES [J].
FEINSTEIN, AR ;
CICCHETTI, DV .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 1990, 43 (06) :543-549
[5]  
GUGENMOOSHOLZMA.I, 1993, STAT MED, V12, P2191
[6]   The meaning of kappa: Probabilistic concepts of reliability and validity revisited [J].
GuggenmoosHolzmann, I .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 1996, 49 (07) :775-782
[7]   Behavior and interpretation of the kappa statistic: Resolution of the two paradoxes [J].
Lantz, CA ;
Nebenzahl, E .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 1996, 49 (04) :431-434
[8]   Pregnancy rate following normal versus abnormal hysterosalpingography findings: A meta-analysis [J].
Maas, JWM ;
Evers, JLH ;
terRiet, G ;
Kessels, AGH .
GYNECOLOGIC AND OBSTETRIC INVESTIGATION, 1997, 43 (02) :79-83
[9]  
MCCLURE M, 1987, AM J EPIDEMIOL, V126, P161
[10]   A METAANALYTIC METHOD FOR SUMMARIZING DIAGNOSTIC-TEST PERFORMANCES - RECEIVER-OPERATING-CHARACTERISTIC SUMMARY POINT ESTIMATES [J].
MIDGETTE, AS ;
STUKEL, TA ;
LITTENBERG, B .
MEDICAL DECISION MAKING, 1993, 13 (03) :253-257