Stability of Jackknife Variance Estimates for Prescription Count Estimates over Time Intervals

• Activity Number: 465
• Type: Contributed
• Date/Time: Wednesday, August 9, 2006 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Christina Gaughan*+ and Heather Zuleba and Chris Boardman and Kennon Copeland
• Companies: IMS Health and IMS Health and IMS Health and IMS Health
• Keywords: jackknife ; survey methodology ; variance estimation

Abstract

In order to obtain variance estimates for product specific point estimates of retail filled prescription (Rx) counts at the national, territory, and precriber level, a jackknife methodology is utilized. The sample data used for jackknife variance estimates are for 400-some products obtained from roughly 70 data suppliers representing approximately 35,000 retail pharmacies. Because variance estimates are required for multiple time intervals (weekly, monthly, and quarterly), 14 weeks of data will be used. These 14 weeks allow for the unique opportunity to test stability of the variance estimates among product specific Rx counts over the time period. The Rx counts for the products will have varying trends over the time period. The relationship of the size and trends of the Rx counts to the stability of variance estimates will be explored.