. Author manuscript; available in PMC: 2010 Jun 1.

Published in final edited form as: Psychol Assess. 2009 Jun;21(2):235–239. doi: 10.1037/a0015686

Table 3.

Estimated prevalence of past-year substance abuse disorders in 1,125 HIV-positive patients, adjusted for measurement error.

Assumption about SCID sensitivity & specificity	Estimated true prevalence (95% CI)
Assuming no error in SCID
100% & 100%	21% (19-23%)
Using latent class analysis estimates
73% & 98%	27% (24-29%)
Varying sensitivity
89% & 98%	22% (19-24%)
56% & 98%	35% (32-38%)
Varying specificity
73% & 96%	25% (22-27%)
73% & 100%	29% (26-31%)
Monte Carlo simulation: Simultaneously considering range of possible values for sensitivity and specificity
(56-89%) & (96-100%)	27% (22-33%)^*

95% simulation interval incorporates both uncertainty associated with random sampling error (as with a standard confidence interval) as well as uncertainty associated with the unknown values of SCID sensitivity and specificity.