* Comparing Conditional Probabilities Just like In ANOVA when we reject Ho for a TEST For Independence , we want To know what The dependence is . To do this we compare pile - pile (or pill - payj') - The conditional probabilities or " For different groups ; and j' ( a j for i and!' ). The primary method used is similar To Tukeys HSD but modified for proportions Ler Zuli be The TesI-Statistic comparing ! conditional on ( Where pili = nis / no p = ( nij + nig ) / ( ne. + new. ) Now, we find The permutation values we will use For our CUTORES . Srep () Calculate all " g" of The observed Zij's .Step in) Find a random permutation , and calculate all " q values of The figs based on The iTh permutation Ler Qi= Ze Calculate Qi. STepici Repear iid By 2000 Times Calculate q ( 2 1 = ( 1 - 2) 100Th percentile or all Qi's . Compare all Zi OBS IO g * ( d ) . _ F Z us > q * ( 2 ). The CBS proportions used in to are sig life