Patrick Dattalo - Böcker

A researcher's decision about the sample to draw in a study may have an enormous impact on the results, and it rests on numerous statistical and practical considerations that can be difficult to juggle. Computer programs help, but no single software package exists that allows researchers to determine sample size across all statistical procedures. This pocket guide shows social work students, educators, and researchers how to prevent some of the mistakes that would result from a wrong sample size decision by describing and critiquing four main approaches to determining sample size. In concise, example-rich chapters, Dattalo covers sample-size determination using power analysis, confidence intervals, computer-intensive strategies, and ethical or cost considerations, as well as techniques for advanced and emerging statistical strategies such as structural equation modeling, multilevel analysis, repeated measures MANOVA and repeated measures ANOVA. He also offers strategies for mitigating pressures to increase sample size when doing so may not be feasible. Whether as an introduction to the process for students or as a refresher for experienced researchers, this practical guide is a perfect overview of a crucial but often overlooked step in empirical social work research.

Random sampling and random assignment are considered by many researchers to be the definitive methodological procedures for maximizing external and internal validity. However, there is a daunting list of legal, ethical, and practical barriers to implementing random sampling and random assignment. While there are no easy ways to overcome these barriers, social workers should seek and utilize strategies that minimize sampling and assignment bias. These methodological and statistical strategies form the book's core. In step-by-step chapters liberally illustrated with examples using a variety of software packages, Dattalo guides readers in selecting and implementing an appropriate strategy. Readers will gain confidence in using such techniques as exemplar sampling, sequential sampling, randomization tests, multiple imputation, mean-score logistic regression, partial randomization, constructed comparison groups, instrumental variables methods, and propensity scores. Each approach will be cataloged in such a way as to highlight its underlying assumptions, implementation strategies, and strengths and weaknesses. Screen shots, annotated resources, and a companion website make this a valuable tool for students, teachers, and researchers seeking a single source that provides a diverse set of tools that will maximize a study's validity when random sampling and random assignment are neither possible nor practical.