This is a problem from Categorical Analysis Book By Alan Agresti

16.18 In an I x J' contingency table, let 9,-1- denote local odds ratio (2.10). a. Show that asymp. cov( log 65;}. V/Tz log f+|.j) : [JTI-;l]'j + J'Cflle]. b. Show that asymp. cov( log {3:}. log f+l.j+l) : \"lm-H\" c. When 63;} and 9\" use mutually exclusive sets of cells, show that asymp. cov( loggj. log 9%) = 0