With a focus on statistical and epidemiological applications, the authors present sample size calculation methods for various study designs ranging from simple two-group comparison assuming normal distribution to complicated multilevel designs involving linear and nonlinear models. The book also combines existing results with R software applications to implement the discussed methods. Sample size estimation is an important, but often, challenging aspect of designing a research study as it involves interrelation between number of subjects (n), significance level ( ), effect-size (ES) and power (1- ). When sample size is under-estimated, studies lack sufficient power to detect statistically significant differences, and as a result, important scientific discoveries can be missed. Alternatively, over-estimation of sample size leads to wastage of time, money and man-power. Conclusions drawn from over-powered studies can be statistically significant, but clinically meaningless, so it is necessary to estimate appropriate sample sizes to correctly identify significance when it exists without wasting valuable resources. The book presents an overview of related mathematical concepts for statisticians as well as the theoretical background for power and sample size calculation. In addition, comprehensive coverage of existing methodologies for sample size and power analysis is provided. The authors review existing software for sample size calculation and utilize R for sample size calculations. Extensive real-life examples demonstrate the examples illustrate the power characteristics of sample size methods.
- ISBN10 1119005698
- ISBN13 9781119005698
- Publish Date 5 May 2015
- Publish Status Active
- Publish Country US
- Publisher John Wiley and Sons Ltd
- Imprint Wiley-Blackwell
- Format Hardcover
- Pages 448
- Language English