x Preface The use of R, a free software environment for statistical computing and graph- ics, throughout the text. Many books claim to integrate technology, but often technology appears to be more of an afterthought. In this book, topics are selected, ordered, and discussed in light of the current practice in statistics, where computers are an indispensable tool, not an occasional add-on. R was chosen because it is both powerful and available. Its “market share” is increasing rapidly, so experience with R is likely to serve students well in their future careers in industry or academics. A large collection of add-on packages are available, and new statistical methods are often available in R before they are available anywhere else. R is open source and is available at the Comprehensive R Archive Network (CRAN, http://cran.r-project.org) for a wide variety of computing plat- forms at no cost. This allows students to obtain the software for their personal computers an essential ingredient if computation is to be used throughout the course. The R code in this book was executed on a 2.66 GHz Intel Core 2 Duo MacBook Pro running OS X (version 10.5.8) and the current version of R (ver- sion 2.12). Results using a different computing platform or different version of R should be similar. An emphasis on practical statistical reasoning. The idea of a statistical study is introduced early on using Fisher’s famous example of the lady tasting tea. Numerical and graphical summaries of data are introduced early to give students experience with R and to allow them to begin formulating statistical questions about data sets even before formal inference is available to help answer those questions. Probability for statistics. One model for the undergraduate mathematical statistics sequence presents a semester of probability followed by a semester of statistics. In this book, I take a different approach and get to statistics early, developing the neces- sary probability as we go along, motivated by questions that are primarily statistical. Hypothesis testing is introduced almost immediately, and p-value computation becomes a motivation for several probability distributions. The binomial test and Fisher’s exact test are introduced formally early on, for ex- ample. Where possible, distributions are presented as statistical models first, and their properties (including the probability mass function or probability density function) derived, rather than the other way around. Joint distribu- tions are motivated by the desire to learn about the sampling distribution of a sample mean. Confidence intervals and inference for means based on t-distributions must wait until a bit more machinery has been developed, but my intention is that a student who only takes the first semester of a two-semester sequence will have a solid understanding of inference for one variable either quantitative or categorical.
Previous Page Next Page