Correction

‘Assessing surrogates as trial endpoints using mixed models’, by Edward L. Korn, Paul S. Albert and Lisa M. McShane ( Statistics in Medicine 2005; 24 :163–182). On page 167, the formulas for R   trial 2and its population quantity are given by $R_{\rm trial}^{2}=(2\hat{\beta}^{2}\hat{\lambda}^{2})/(2\hat{\beta}^{2}\hat{\lambda}^{2}+\hat{\gamma}{2}) \,{\rm and} \,(2 {\beta}^{2} {\lambda}^{2})/(2 {\beta}^{2} {\lambda}^{2}+{\gamma}^{2}),$ respectively. However, since the variance of g   ij ( D )is 2γ 2 , the correct formulas are $R_{\rm trial}^{2}=(2\hat{\beta}^{2}\hat{\lambda}^{2})/(2\hat{\beta}^{2}\hat{\lambda}^{2}+2 \hat{\gamma}{2}) \,{\rm and} \,(2 {\beta}^{2} {\lambda}^{2})/(2 {\beta}^{2} {\lambda}^{2}+2 {\gamma}^{2}),$ respectively. The only place these formulas are used in the paper is in the example in Section 3; for that example $\hat {\gamma}^2=0$ , so that the results are unaffected.