Figure 8: Lineup to examine the signi_cance of the iron e_ect on RPA (green), regardless of the e_ect on EV (orange). The question at the top of the lineup is the request put to the judges. One of the plots uses the real data, and the true experimental design, the rest use permuted treatment levels e_ectively breaking any real association between experimental factors and gene expression. In each case the gene shown is the one that has the most signi_cant di_erence between iron response on RPA. The position of the observed data plot in the lineup is the solution of this expression