Measuring the Change in Contingency Tables Using Markov Models — Application to the Effect of Preceding Conception on the Next one
A. K. Saleh, Mohamed A.A. Moussa, H. Hathout, R. A. Kasraoui · Methods of Information in Medicine · 1982
A method is described for the analysis of time sequence data viewed as contingency tables. This includes 1.) determining the maximum likelihood estimates of stationary and transition probabilities; 2.) testing stationarity of transition probabilities in first and second order Markov models by fitting the data to the appropriate configuration of log-linear models; 3.) assessing the order of dependence through fitting log-linear models. Data based on a study recently conducted on women delivered in the Principal Maternity Hospital in Kuwait is used for illustration. It is concluded that conception outcome depends on pregnancy order as well as on at least two previous pregnancy outcomes.