Results

**1 - 2**of**2**### Table 1. Possible outcomes of individual tests. Note that V is the number of type I errors; T is the number of type II errors; and only m, r and mC28/r are observed while the other variables are unknown.

2004

"... In PAGE 4: ... The false non discovery rate (FNR) A natural companion to the FDR is the false non discovery rate (FNR), or the expected proportion of non rejections that are incorrect (Genovese and Wasserman 2002). This is the ratio T/mC28/r in Table1 . The FNR plays a similar role in FDR control as power does in FWER control: given a procedure to decide which p values are significant and which are not, it quantifies a rate of making type II errors.... In PAGE 4: ...002). This is the ratio T/mC28/r in Table 1. The FNR plays a similar role in FDR control as power does in FWER control: given a procedure to decide which p values are significant and which are not, it quantifies a rate of making type II errors. The FNR can be estimated based on an estimate of the proportion of true null cases (m0/m in Table1 ) combined with an expecta- tion for the number of false discoveries (from an FDR controlling procedure; V/r in Table 1, Genovese and Wasserman 2002, Taylor et al. in press).... In PAGE 4: ...002). This is the ratio T/mC28/r in Table 1. The FNR plays a similar role in FDR control as power does in FWER control: given a procedure to decide which p values are significant and which are not, it quantifies a rate of making type II errors. The FNR can be estimated based on an estimate of the proportion of true null cases (m0/m in Table 1) combined with an expecta- tion for the number of false discoveries (from an FDR controlling procedure; V/r in Table1 , Genovese and Wasserman 2002, Taylor et al. in press).... ..."

### Table 1: Possible Outcomes from m Hypothesis Tests

"... In PAGE 3: ... Table1 displays the various outcomes when testing m genes. The quantity V is the number of false positives (Type I errors), while R is the total number... In PAGE 6: ... In our example, if we take c = 2:9 , we get R = 609, ^ V = 31:9, ^ 0 = :70, giving [ FDR = :037 3 The Miss Rate Having derived a list of genes from a using a rule like jTij gt; c, it is of interest to estimate some sort of false negative rate. Looking at Table1 , the quantity E(Q=W ) is what Genovese amp; Wasserman (2003) call the false non-discovery rate. This quantity is the proportion of false negatives among all genes with jTij lt; c.... ..."

**1 - 2**of

**2**