High survey nonresponse in unemployment duration studies may have a strong effect Ashley Wilde on inference if exit from unemployment affects the chance of nonresponse (nonresponse causality).In rotational studies large part of the nonresponse results from panel attritions.A method to test the presence of the causality mechanism for rotating panel designs is proposed and its asymptotic Support Socks consistency is proved under the Cox regression model.An application to real labor data and a simulation study are shown.