No label defined (Q3341650)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3876376
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 3876376

      Statements

      0 references
      1981
      0 references
      comparison problems
      0 references
      graphical method
      0 references
      box plots
      0 references
      Two-way tables
      0 references
      categorical variable
      0 references
      association
      0 references
      robustness
      0 references
      Large sample confidence intervals
      0 references
      experimental design
      0 references
      Neyman-Pearson theory
      0 references
      sensitivity measures
      0 references
      power curve
      0 references
      small sample case
      0 references
      Student's t-distribution
      0 references
      trimmed mean
      0 references
      transformation of variables
      0 references
      Wilcoxon two-sample test
      0 references
      K- sample problem
      0 references
      frequency distributions
      0 references
      paired difference method
      0 references
      randomized block design
      0 references
      two sample t-method
      0 references
      signed-rank test
      0 references
      chi- square goodness-of-fit test
      0 references
      two-way tables
      0 references
      multiple-comparison procedures
      0 references
      homogeneity
      0 references
      contingency tests
      0 references
      linear regression
      0 references
      linear models
      0 references
      least squares
      0 references
      residuals
      0 references
      error bounds for linear predictions
      0 references
      outliers
      0 references
      According to the author this book was designed in the early 1970's to serve as a beginning course in several departments at the University of New Mexico - psychology, sociology, economics, political sciences and business. The aim of the book is to make the students fit to perform and interpret the basic methods of inference on data as well as to be able to read and understand the applications of statistics in their fields. The author underlines that though there are limits to the usefulness of hand analysis, a large core of the most important methods in statistics can be conveniently applied by hand to data sets of sizes commonly seen in practice. In the first chapter random variables and their classification are considered and useful methods for describing frequency distributions are given, that are partly graphical, but have, in addition, other features that make them convenient for rapid exploratory analyses of data sets. In the second chapter measures for frequency distributions are described. Two location parameters, mean and median and measures of dispersion associated with them are introduced together with their sample versions. A graphical display called the box plot which is a powerful tool for comparing distributions is constructed. The third chapter provides an exploration of comparison problems. A useful graphical method is given for making measurement variable comparisons, which employs the use of box plots. Two-way tables for organizing categorical variable comparisons are provided along with a useful symbolic technique for pinpointing interesting distributional differences. The association problem involves the relationship between two (or more) variables in a population. An exploratory treatment of this problem is the subject of chapter 4. The goal of chapter 5 is to provide some general ideas about the nature of probability and probability models and to introduce some probability notations and terminology that will be used later in the book. In the sixth chapter the general concept of distribution of a measurement variable is introduced and illustrated by the normal distribution. The reader becomes familiar with the probability notation for measurement variables and learns to use the tables of normal distributions. Specific sampling distributions needed in chapters 8 and 9 are given in chapter 7, along with a discussion of how and why they enter the statistical scene. The important concept of standard error of an estimator is also introduced and illustrated. In chapter 8 the study of statistical inference begins with the topic of confidence intervals. The application to the selection of sample size is illustrated for a variety of problems. The concept of robustness of statistical procedures is introduced. Large sample confidence intervals are constructed for the mean \(\mu\) and the parameter \(\rho\) of a bivariate variable. An alternative confidence interval for \(\mu\), which performs better than the standard confidence interval in many cases, is given. In chapter 9, hypothesis testing is studied. The P-value is introduced in continuation of the discussion of evidence about the hypothesis, first presented in chapter 7. P-values are derived for tests involving the parameters \(\rho\) and \(\mu\) for large sample sizes. The experimental design aspect of this procedure is accounted for by the introduction of the Neyman-Pearson theory. The use of hypothesis-testing sensitivity measures and the power curve, for designing sample size is introduced. Finally, the P-value is linked to the decision method to form a cohesive and practical theory. The tests and confidence intervals for the mean of a measurable variable in the small sample case, considered in chapter 10, differ in one basic respect from those given in chapters 8 and 9 - namely, in the sampling distribution for Student's t-ratio. The key assumption under which Student's t-distribution is the sampling distribution of the one- and two-sample statistics is that the observations be normally distributed. When this assumption is not valid the t-tests and confidence intervals can suffer a substantial loss of sensitivity. One treatment of this problem discussed here involves the use of a robust estimator such as the trimmed mean. Another is based on the idea of a transformation of variables whereby the distribution of observations is modified to conform better with the assumption of normality. One such transformation leads to the Wilcoxon two-sample test. A version of this test is given in this chapter. In the K-sample problem treated in chapter 11, an attempt is made to detect differences in the frequency distributions of a measurement variable for \(K=3\), or more populations based on random samples from each population. As in chapter 10, the comparison is based on a location measure such as the population mean. The title of chapter 12 is ''Experimental design and the paired difference method''. The key concepts and methods of this chapter are as follows: the paired-difference method; experimental set-up that leads to the use of a randomized block design; paired-difference method versus the two sample t-method; trimmed-mean paired differences; signed-rank test for paired differences; signed rank test for the one-sample problem. In chapter 13, the author returns to the categorical variables comparison and association problems treated in an exploratory fashion in chapters 3 and 4. Here are considered the following key concepts and methods: scientific hypothesis testing; the chi-square goodness-of-fit test; the chi-square test for homogeneity and contingency test; tests of equality of population proportions based on two-way tables; confidence intervals for the difference of population proportions; multiple-comparison procedures for homogeneity and contingency tests. The last two chapters are devoted to linear regression. In chapter 14 descriptive methods are given. The main topics of this chapter are as follows: the linear regression model; linear models and least squares; fitting the regression line by least squares; a descriptive measure of linearity: the coefficient of determination; use of residuals for the study of regression curve linearity. In the final fifteenth chapter statistical inference results are considered for problems involving linear regression. These include confidence intervals for the slope parameter \(\beta_ 1\) and for the regression line, and error bounds for linear predictions. The interpretation of the usual hypothesis test for \(H_ 0:\beta_ 1=0\) is also discussed. Finally, the influence of outliers in linear regression is discussed and a simple strategy to diagnose the presence of this influence is given.
      0 references

      Identifiers