Coherence Structures of Mortality Time Series Data for All Causes of Deaths for Selected Neighboring States 2004--2006

• Activity Number: 409
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Diba Khan*+ and Benjamin Kedem and Myron Katzoff
• Companies: National Center for Health Statistics and National Center for Health Statistics and National Center for Health Statistics (retired)
• Keywords: mortality ; coherence

Abstract

The National Center for Health Statistics collects vital statistics for all states including mortality data from all causes of deaths. The data used in the analysis are weekly mortality counts that cover a span of 3 years from all causes of deaths from neighboring states. For prediction and regression purposes it is important to find relationships between these weekly time series. A quantity which measures linear relationships between time series is the measure of coherence. A certain nonlinear extension of linear coherence is called lagged coherence. Both measures were applied to selected neighboring states and they indicate strong associations between neighboring states particularly at low frequencies or for long cycles. This helped in identifying useful covariates for prediction and regression models for mortality time series. This paper illustrates some prediction and regression models.