THE USE OF CLASSICAL STATISTICS IN DERIVING AND EVALUATING CER'S
Abstract
Regression theory is frequently used in the development of cost estimating relationships (CERs). Unfortunately there is a tendency to use this tool and the statistics that are associated with it without fully understanding either. A good understanding, however, is necessary for both the user and the reviewer. This report considers two main topics. The first centers on the meaning of some of the commonly used statistics and the differences between common interval estimates. The second addresses the applicability of the usual interpretation of these statistics and interval estimates in cost analysis and the possible meanings that might be attached to them even if statistical assumptions are not fully satisfied. This presentation will address the above through a discussion of the following topics: (1) Assumptions of the multiple linear regression model and how well they are fulfilled in the cost analysis application. (2) Least squares estimators as 'best' estimators. (3) Properties of some commonly used statistics from a geometrical point of view. (4) Differences in commonly used interval estimates.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 17, 1966
- Accession Number
- AD0659323
Entities
People
- Charles Graver
Organizations
- United States Assistant Secretary of Defense