Sense and Nonsense of Statistical Inference: Controversy: Misuse, and Subtlety

ISBN-13: 9780824787981 / Angielski / Twarda / 1992 / 256 str.

Chamont Wang; Wei Wang; Charmont Wang

Opis

This volume focuses on the abuse of statistical inference in scientific and statistical literature, as well as in a variety of other sources, presenting examples of misused statistics to show that many scientists and statisticians are unaware of, or unwilling to challenge the chaotic state of statistical practices. biomedical and behavioural sciences, economics and the statistical literature; discusses conflicting views of randomization, emphasizing certain aspects of induction and epistemology; reveals fallacious practices in statistical causal inference, stressing the misuse of regression models and time-series analysis as instant formulas to draw causal relationships; treats constructive uses of statistics, such as a modern version of Fisher's puzzle, Bayesian analysis, Shewhart control chart, descriptive statistics, chi-square test, nonlinear modeling, spectral estimation and Markov processes in quality control.

Recenzje

". . .Excellent guide for statistical users, educators, students. Shows that improper use of statistical inference can do more harm than good. "
---American Mathematical Monthly
". . .very informative. . . . . .should be read by all users of statistics. "
---Journal of Quality Technology
". . .illuminating, entertaining. . . . . .this book will be found most useful by graduate users of statistics who have been taught the techniques but have not subject them to critical scrutiny. . .professional statisticians will find fresh ideas about their subject here. "
---Australian Journal of Statistics

Spis treści

Part 1 Fads and fallacies in hypothesis testing: examples - the t-test; a two-stage test-of-significance; more examples - a Kolmogorov-Smirnov test; mechanical application of statistical tests; data snooping; an appreciation of non-significant results; type I and type II errors - for decision making; type I and type II errors - for general scientists. Part 2 Quasi-inferential statistics: randomness or chaos? Hume's problem; unobservables, semi-unobservables and grab sets; is statistics a science?; grab sets and quasi-inferential statistics; concluding remarks - quasi- and pseudo-inferential statistics. Part 3 Statistical causality and law-like relationships: sense and nonsense in causal inference - examples; Rubin's model and controlled experiments; Rubin's model and observational studies; causal inference in sample survey and other observational studies; causes, indicators and latent variables. Part 4 Amoeba regression and time-series models: discovering causal structure - science now can be easily cloned; regression and time-series analysis - science or lunacy? (part I); regression and time-series analysis - science or lunacy? (part II); regression and time-series analysis - science or lunacy? (part III); statistical correlation versus physical causation. Part 5 A critical eye and an appreciative mind toward subjective knowledge: the sorry state of statistical evaluation - a case study in educational research; modeling interaction effects - a case study from the social-behavioural sciences. Part 6 On objectivity, subjectivity and probability: statistical justification of scientific knowledge and scientific philosophy; classical probability, common sense and a strange view of nature; intuition and subjective knowledge in action - the Bayes theorem (and its misuse); Bayesian time-series analysis and E. T. (extra time-series) judgment; a pursuit of information beyond the data in randomized and nonrandomized studies; women and love - a case study in qualitative/quantitative analysis. Part 7 A delicate balance between order and chaos: reliability; order within chaos - heartbeats, brainwaves, simulated annealing and the fiddling of a system. Part 8 The riddle of the ubiquitous statistics: information and misinformation; statistical quality control and the conflicting teachings of Q.C. Gurus. Epilogue - toward a new perspective on statistical inference.

O autorze

Wang\, Charmont