MI works there too. As an example, check out Hopke, Liu,and Rubin Biometrics, 2001, p.22, where there are a couple of time series models used (multivariate too), with both fully missing values and below threshold values allowed.
On Fri, 15 Nov 2002, Gerardo Hurtado wrote: > Does anyone have references about imputation methods in time series > models?....or if Multiple Imputation is not the methodology to handle > missing values, does anyone have references about missing data in Time > Series models? > > Thank your for your time. > > Gerardo Hurtado > SAS Institute > Ph: 919-531-7163 > -- Donald B. Rubin John L. Loeb Professor of Statistics Chairman Department of Statistics Harvard University Cambridge MA 02138 Tel: 617-495-5498 Fax: 617-496-8057
