Cost-effective Design of Growth Studies with Aggregation and Tracking

Journal of Biometrics & Biostatistics

article

Shu Jiang¹  Richard J Cook¹ 

[1] Department of Statistics and Actuarial Science, University of Waterloo

关键词: Aggregate data;    Clustered data;    Design;    Heterogeneity;    Interval censoring;    Markov process;    Multistate model;   

DOI  :  10.4172/2155-6180.1000406

来源: Hilaris Publisher

【 摘 要 】

Studies of the development and growth of organisms are often conducted in laboratories where organisms maintainedin tanks are examined repeatedly over time. Collection and recording of cross-sectional aggregate count data on stageoccupancy is both less expensive and administratively more convenient than tracking the stages of each organism overtime. In such settings tank to tank variation must also be taken into account as growth rates may be more similar amongorganisms within the same tank than for those in different tanks. We consider the cost effect design of a prospectivedevelopmental study of organisms based on a marginal Markov model which deals with between tank variation and withintank dependence. We develop a flexible design in which some tanks provide repeated cross-sectional aggregate data,and other tanks provide serial responses through tracking individuals. We assess the relative efficiency of aggregate andindividual-level longitudinal data. The optimal cost-effective design is shown to depend on whether primary interest lies intransition intensities or associated cluster-level covariate effects.

【 授权许可】

Unknown