Catálogo de publicaciones - libros

In this chapter, you learned and quickly mastered the use of the primitive or percentile bootstrap to obtain estimates of the precision of estimates. You then learned a variety of methods including the BC_α bootstrap, the bootstrap-t, the tilted bootstrap, the iterative bootstrap, and bootstrap smoothing for improving the accuracy of confidence intervals. Computer code was provided to aid in putting these methods into practice.

In this chapter, you learned that power, sample size, and significance level are interrelated. You learned that your choice of test statistic will depend on the hypothesis, the alternative, the loss function, and the type of test you employ. You learned the value of exact unbiased tests. You learned the assumptions underlying and the differences among test of hypotheses based on the bootstrap, the permutation test, and parametric distributions.

In this chapter, you learned the principles of experimental design: to block or measure all factors under your control, to randomize with regard to factors that are not. You learned to analyze balanced k -way designs for main effects, and balanced two-by-two designs for both main effects and interactions. You learned to use the Latin Square to reduce sample size and to use bootstrap methods when designs are not balanced.

In this chapter, you were introduced to the concept of a contingency table with fixed marginals, and shown you could test against a wide variety of general and specific alternatives by examining the resampling distribution of the appropriate test statistic. Among the test statistics you considered were Fisher’s Exact, Freedman-Halton, Pearson’s Chi-Square, Tau, Q, Pitman’s correlation, and linear-by-linear association. These latter two statistics are to be used when you can take advantage of an ordering among the categories.

In this chapter, you learned the essentials of multivariate analysis for two-sample comparisons and applied them to repeated measures on the same subject. You learned how to detect clustering in time and space and to validate clustering models. You used the generalized quadratic form in its several guises including Mantel’s U and Mielke’s multiresponse permutation procedure (MRPP) to work through applications in archaeology, epidemiology, and ornithology. And you learned how to combine the results of multiple, simultaneous analyses.

In this chapter, you reviewed the steps in model development beginning with descriptive statistics such as scatter plots and correlations to identify relationships among variables. You also considered some of the possible caveats. You learned a number of methods for ascertaining the statistical significance of model coefficients and for deriving the associated confidence intervals. You learned it is essential to distinguish between goodness of fit and prediction, and were provided with a variety of cross-validation resampling methods to tie the two together.