Method agreement analysis: A review of correct methodology

Fagan's nomogram. (Adapted from Sackett Dl, Richardson WS, Rosenberg W, Haynes RB. Evidence-based Medicine: How to Practice and Teach EBM, 2nd Edn., Churchill-Livingstone, 1977, with permission.)

Receive operating characteristic curves illustrating a poorly discriminating test (blue circles, AUROC 0.517) and a test that has good discriminating power (red circles, AUROC 0.956; green circle indicates the point of greatest discrimination). (Redrawn from Martínez-Pastor F, del Rocío Fernández-Santos M, Domínguez-Rebolledo ÁE, Esteso MC, Garde JJ. DNA status on thawed semen from fighting bull: a comparison between the SCD and the SCSA tests. Reprod Domest Anim 2008;44:424–431.)

Bland and Altman diagram showing the plot of the difference between the diameters (mm) of the equine follicle just prior to ovulation in two consecutive cycles of the mare against the mean of the pair (n=20). Red lines show limits of agreement, and the purple line shows the mean value of the differences. The green line is the zero line used to assess the discrepancy of the observed mean difference from zero. (Data from Ref. 10, courtesy of Dr. Cuervo-Arango.)

Bland and Altman plot showing limits of agreement between two methods of measuring sperm motility, an objective Hamilton-Thorne computer-based semen analyzer and a subjective visual assessment, of samples of boar semen. is the mean difference, and s_d is the standard deviation of the differences between pairs of measurements. (Redrawn and modified from Vyt P,Maes D, Rijsselaere T, Dejonkheere E, Castryck F, Van Soom A. Motility assessment of porcine spermatozoa: a comparison of methods. Reprod Dom Anim 2004;39:447.)

Diagram showing two kinds of association between the results of Method 1 and those of Method 2. The red circles on the upper line demonstrate perfect correlation but no agreement. The blue circles around the lower line demonstrate poor correlation but no systematic difference between the two methods.

A scatterplot of the diameter of the ovulating follicle just prior to ovulation from two consecutive cycles in 20 mares [10]. The line of best fit is drawn though the points.