Kulinskaya E., Morgenthaler S., Staudte R.G. Meta Analysis: A Guide to Calibrating and Combining Statistical Evidence

Wiley, 2008. — 272 p. — ISBN: 0470028645, 9780470028643Meta Analysis: A Guide to Calibrating and Combining Statistical Evidence acts as a source of basic methods for scientists wanting to combine evidence from different experiments. The authors aim to promote a deeper understanding of the notion of statistical evidence.
The book is comprised of two parts The Handbook, and The Theory. The Handbook is a guide for combining and interpreting experimental evidence to solve standard statistical problems. This section allows someone with a rudimentary knowledge in general statistics to apply the methods. The Theory provides the motivation, theory and results of simulation experiments to justify the methodology.
This is a coherent introduction to the statistical concepts required to understand the authors thesis that evidence in a test statistic can often be calibrated when transformed to the right scale.The Methods.
What can the reader expect from this book?
Independent measurements with known precision.
Independent measurements with unknown precision.
Comparing treatment to control.
Comparing K treatments.
Evaluating risks.
Comparing risks.
Evaluating Poisson rates.
Comparing Poisson rates.
Goodness-of-fit testing.
Evidence for heterogeneity of effects and transformed effects.
Combining evidence: fixed standardized effects model.
Combining evidence: random standardized effects mode.
Meta-regression.
Accounting for publication bias.
The Theory.
Calibrating evidence in a test.
The basics of variance stabilizing transformations.
One-sample binomial tests.
Two-sample binomial tests.
Defining evidence in t-statistics.
Two-sample comparisons.
Evidence in the chi-squared statistic.
Evidence in F-tests.
Evidence in Cochran’s Q for heterogeneity of effects.
Combining evidence from K studies.
Correcting for publication bias.
Large-sample properties of variance stabilizing transformations.