Notes for developers of selection procedures: An addendum to: J. Branke, S.E. Chick, C. Schmidt, 2007, Selecting a Selection Procedure (including Online Companion), Management Science, 53(12): 1916-1932.

1. The expected value of information procedures are directly implementable as described in the paper.

2. The OCBA variations may need additional description.

a. Equation (9) gives the formula for the EAPCS_i when the best system is defined by the lowest expected mean output.  To be consistent with the other sections, remove the "1 -" in Equation (9) to obtain the EAPCS_i when the best system is defined by the largest expected mean output.

b. Equation (9) also refers to $\tilde\nu_(j)(k)$, the approximate posterior degrees of freedom for two random variables $W_(j)$ and $W_(k)$ with student distribution with Welch's approximation.  The value of $\tilde\nu_(j)(k)$ is defined by analogy with Equation (4), which defines $\nu_(j)(k)$.  To obtain the value of $\tilde\nu_(j)(k)$, substitute $n_(j)$ and $n_(k)$ in Equation (4) with the total number of replications with the replications to be observed, $n_(j) + \tau_(j)$ and $n_(k) + \tau_(k)$, respectively, where $\tau_i$ is the number of additional replications to be allocated to system $i$.

c. The use of the approximate degrees of freedom might not even be necessary - it adds computational effort but does not necessarily improve performance (e.g., see Chen, C.-H., E. Yücesan, L. Dai, and H. Chen. 2007. Efficient computation of optimal budget allocation for discrete event simulation experiment. IIE Transactions:to appear., or Branke, Chick and Schmidt, WSC 05).