Some properties of a moment estimator for the index parameter of negative binomial distribution

Conference · Fri Jan 01 00:00:00 EST 1988

OSTI ID:7153582

Lam, H K; Shenton, L R; Bowman, K O

Taking the mean of the negative binomial to be kP and the variance kP(1+p), a moment estimator of k is (m/sub 1/')/sup 2//(m/sub 2/-m/sub 1/'). From previous studies it is known that the series development for the moments of the momentor estimator k* appear to be markedly divergent (probably due to singularity in the denominator). The present study shows that the moments of 1/k* are more stable than those of k*. Thus, if we wish to make inference about k by using 4-moment distribution fitting, then the moments of 1/k* should be used, rather than those of k*. 2 refs., 2 tabs.

• Research Organization: Bell Communications Research, Inc., Red Bank, NJ (USA); Georgia Univ., Athens (USA); Oak Ridge National Lab., TN (USA)
• DOE Contract Number: AC05-84OR21400
• OSTI ID: 7153582
• Report Number(s): CONF-880872-3; ON: DE88011436
• Resource Relation: Conference: Joint statistical meetings, New Orleans, LA, USA, 22 Aug 1988; Other Information: Portions of this document are illegible in microfiche products
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
STATISTICS
ANALYTICAL SOLUTION
CALCULATION METHODS
DISTRIBUTION FUNCTIONS
EQUATIONS
MATHEMATICAL MODELS
MATHEMATICS
PERTURBATION THEORY
FUNCTIONS
990230* - Mathematics & Mathematical Models- (1987-1989)