We present a method for estimation of the parameters of an AR(1) process with missing data. This method uses multiple imputation to permit estimation of the parameters of the process and their variances as accurately as possible. I find that there is an inevitable obstacle for a good imputation method for autoregressive processes. This obstacle arises because of the correlation in the data, which causes measurement error bias.
Supervisor: Paul Shaman. Thesis (Ph.D. in Statistics) -- University of Pennsylvania, 2001. Includes bibliographical references.