scientific article; zbMATH DE number 2051800

MaRDI QIDQ4453851zbMATH OpenFDO

Authors C. Matthew Jones

Publication date 7 March 2004

zbMATH Keywords

harmonic mean of Poisson distributionsmulti-center clinical trialsnegative momentsPoisson distributionPoisson-Charlie polynomialssamplingsimultaneous recruitment of patientsTiku estimators

Mathematics Subject Classification ID

Probability distributions: general theory (60E05) Distribution theory (60E99) Approximations to statistical distributions (nonasymptotic) (62E17) Applications of statistics to biology and medical sciences; meta analysis (62P10)

Review text

The paper deals with the problem of estimating negative moments of the Poisson distribution and moments of the harmonic mean of Poisson distributions, in connection to work on randomized multi-center clinical studies. For the negative moments of the so-called positive Poisson random variable with parameter \(\lambda\), the paper derives asymptotic expansions and related approximations for the first four negative moments \(\mu_{-1},\dots,\mu_{-4}\) of the Poisson distribution, when the parameter \(\lambda\to\infty\). The methods for approximating \(\mu_{-1}\) and \(\mu_{-k}\), \(k\geq 2\), are different. For computing \(\mu_{-1}\) a simple recurrence formula is used, and as comparison benchmark the well-known Tiku's estimators are employed. For the derivation of higher-order negative moments \(\mu_{-k}\), \(k = 2, 3, 4\), the computation method is based on the properties of the so-called Poisson-Charlie polynomials. The analytical formulae for some of these approximations are exposed (in the Appendix), and the resulted accuracy is shown to be quite good. A similar methodology is applied for computing the first four moments of the harmonic mean of (two or a finite number of) independent and identically distributed (IID) Poisson random variables. The approximation of the negative moments of the Poisson distribution has attracted a special interest for applications in the field of sampling. Estimating the harmonic mean of IID Poisson distributions is important in connection to work on randomized multi-center clinical trials in the pharmaceutical industry, when enrollment is accelerated by the simultaneous recruitment of patients to numerous centers.

Recommendations

• Approximating the negative moments of the Poisson distribution.
• On the Poisson approximation to the negative hypergeometric distribution
• scientific article; zbMATH DE number 5233602
• Poisson approximation for a sum of negative binomial random variables
• The negative integer moments of the generalized poisson distribution
• scientific article; zbMATH DE number 5305195
• An improved Poisson to approximate the negative binomial distribution
• Approximation of the distribution of some statistics by Poisson laws

Cited in

(1)

• Multi-Center Clinical Trials with Random Enrollment: Theoretical Approximations

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4453851)