More Details. Helena Chmura Kraemer 6 books0. Search review text. Essentially the same review i. The newer version makes for more interesting reading than the original, which is more tersely written; it has bigger pages, more of them, and fewer blank chapter headings.
Thus, the content is about doubled with most of the added text devoted to various pitfalls when designing or carrying out a project. The intended audience is medical researchers and behavioral scientists. A dozen commonly used statistical tests are discussed in a unified manner. When a hypothesis test is conducted at some level of significance, power is the likelihood of rejecting the null hypothesis.
It depends on sample size, the suitability of the test, and how far the true population parameter is from the null value. For each test, the researcher decides on a measure such as how big a difference between the means of two populations is worth reporting.
Then, depending on which test is used, a formula converts the intuitive measure to a "critical effect" known as DELTA from which a required sample size is determined.
The first appendix lists the conversion formulas which employ such familiar functions as powers squares and square roots , quotients, exponentials, and arcsines. I suppose by using weird functions, a similar conversion from an intuitive measure to the common DELTA would be possible for any other test. A set of four tables level of significance. I dare say it takes more than a pedestrian understanding of basic statistical methods to follow the narratives or make use of the tables.
In some instances, successive formulas can be seen to be related by simple algebra steps but none of the formulas are derived from first principles. References are given for some of them. Someone who opens the book thinking he or she will find a sample size recommendation will encounter discussions of many other topics. The narratives relate statistical sampling to the scientific method as a way of adding to knowledge.
Topics include whether to dichotomize values, how many explanatory variables to include in a regression model, biases arising from various sampling methods, advantages of random treatment assignments over an observational study, how much of a pilot study should be published, and a few other issues related to funding a study.
That's just the tip of an iceberg. Some are annoying but most are minor with little harm done. The latter is the Number Needed to Treat to achieve an improvement from using one treatment over another.
View Larger Image. Ask Seller a Question. Title: How Many Subjects? With increased emphasis on helping readers understand the context in which power calculations are done, this Second Edition of How Many Subjects? Because the same technique is used with only slight modifications for different statistical tests, researchers can then easily compare the sample sizes required by different designs and tests to make cost-effective decisions in planning a study.
These comparisons demonstrate important principles of design, measurement, and analysis that are rarely discussed in courses or textbooks, making this book a valuable instructional resource as well as a must-have guide for frequent reference. She has published extensively in the behavioral as well as statistical literature.
In retirement, she continues to serve on several editorial boards, and consult on research projects. Visit Seller's Storefront. List this Seller's Books. Payment Methods accepted by seller.
0コメント