. Author manuscript; available in PMC: 2009 Oct 1.

Published in final edited form as: J Am Stat Assoc. 2008 Mar 1;103(481):61–73. doi: 10.1198/016214507000000329

Table 1.

Estimation of overall prevalence, sensitivity, and specificity for digital radiography using no gold standard (GS) and with the consensus rating as the gold standard

Estimator	Model	GS	No GS
P_d	CI	.24_(.04)^*	.18_(.04)
	GRE	.24_(.04)	.16_(.10)
	FM	.24_(.04)	.17_(.04)
SENS	CI	.75_(.06)	.89_(.05)
	GRE	.75_(.06)	.92_(.19)
	FM	.75_(.06)	.91_(.05)
SPEC	CI	.91_(.01)	.89_(.02)
	GRE	.91_(.01)	.88_(.03)
	FM	.91_(.01)	.90_(.02)

NOTE: Models were fit under the conditional independence (CI), finite mixture (FM), and Gaussian random effects (GRE) models using Iinuma et al.'s data.

Standard errors were estimated using a bootstrap with 1,000 bootstrap samples.