UBC Statistics Department

[ P. Gustafson's Home ]


    [ Faculty Index ]

    [ People Index ]

    [ Statistics Home Page ]

P. Gustafson

Measurement Error and Misclasification in Statistics and Epidemiology: Impacts and Bayesian Adjustments

Measurement Error and Misclassification in Statistics and Epidemiology (MEMSE for short) was published by Chapman and Hall / CRC Press in the Fall of 2003. (It is the 13th volume in their Interdisciplinary Statistics Series.) Check out the publisher's webpage about the book.

Errata for MEMSE (pdf), last updated May 2, 2005.

Some R software and data used in MEMSE is given below. (Last updated July 21, 2004. I apologize that this is still very incomplete. I will do more updates as time permits.)

Ch. 4 Software (Adjusting for Measurement Error)

logreg1() - MCMC for logistic regression with measurement error, supports both normal exposure model and flexible `reverse exposure' model (see Secs. 4.4, 4.5, 4.7 of MEMSE). Also logreg0() fits logistic regression without measurement error.

Ch. 5 Software (Adjusting for Misclassification)

dual.ind() - MCMC for DUAL-IND model: two imperfect exposure assessments applied to two popluations (i.e., control and case), with assumption that the two assessments are conditionally independent given the true exposure status. (See Sec. 5.3 of MEMSE.)

dual.dep() - MCMC for DUAL-DEP model: two imperfect exposure assessments applied to two popluations, without the independence assumption, but with prior distributions on the dependence parameters. (See Sec. 5.3. of MEMSE.)

dual.ind.reg() - MCMC for extension of DUAL-IND model to handle covariates. Contains two functions: dual.ind.reg1() for modelling exposure given outcome and covariates, dual.ind.reg2() for modelling outcome given exposure and covariates. (See Sec. 5.4 of MEMSE.)

[ Return to P. Gustafson's Home Page ]

Statistics:  Faculty Index || People Index || Statistics Home Page 

Contact Information Link Image