- ホーム
- > 洋書
- > 英文書
- > Science / Mathematics
Full Description
Written with the data analyst in mind and based on a course taught at the University of Washington, this volume answers key practical questions professionals typically encounter, such as which data analysis method to use and how to interpret and communicate results. Starting by looking at the desired outcome, this book then works backwards. It gives guidance on ways to communicate the results of analyses, and then covers the formulation of scientific questions, before tackling study design, and statistical modeling with emphasis on ways to deal with assumptions. The book is full of examples that use real data and gives software suggestions in an appendix.
Contents
Part IPapers and Reports The Structure of a Scientific Paper The Background The Methods The Results The Discussion Consulting Reports Writing Style Scientific Presentations The Structure of a Presentation The Background The Methods The Results The Discussion Oral Presentations versus Posters Descriptive Statistics Fundamental Concepts Effective Tables Effective Statistical Graphics Descriptives for Papers Versus Presentations Part II: Analysis by Design Analysis of Associations What is an Observational Study? Formulating the Scientific Question (Confounding) Estimating Associations (stratification versus adjustment) Interpreting Results (Hill's criteria, Selection bias, Information Bias) Analysis for Prediction What is a Predictive Study? Building a Predictive Model Assessing Prediction Error Part III. Using Statistical Models Assumptions about the Response Mean Model Misspecification and its Impact on Inference Model Selection Confounding and Mediation Effect Modification Assumptions about the Response Variance The Impact of Misspecified Variances Accounting for Misspecified Variances: Robust Variance Estimators Modeling the Variance to Increase Precision: Quasi-likelihood Assumptions about the Response Distribution The Impact of Non-normality on Inference in Linear Models The Impact of Misspecified Response Distributions in Generalized Linear Models Using Resampling to Assess Uncertainty References Appendices Index
