Читать книгу The Tao of Statistics - Dana K. Keller - Страница 9

 The world is hazy

 No clean lines or sharp points

 Focus the camera

The world of statistics starts with a question. Many types of statistics exist because questions can be of almost any type, can be about almost anything, and can take place under quite varied conditions. This variety calls for an almost equal number of statistical approaches to answers.

Statistics need data. Data come from measures. Good answers to important questions require good data from relevant measures. Good measures are specific and, through that specificity, are informative. If the selected measures do not directly address the questions, the statistics can be meaningless because they will have lost their context.

Statistics become important when they are about things that people consider important. In this book, we will follow two professionals with different careers as they use statistical concepts to answer questions that are relevant to their own lines of work. The first is a high school principal with questions about differences in quality and other characteristics of the various classes taught at his school. The second is a director of public health with questions about how well the residents of her state are meeting national guidelines for public health issues, such as immunizations, especially for individuals receiving public medical assistance.

The choice of careers and the specific concerns of the related fields really do not matter to the interpretation of statistics. What matters is that appropriate data are used with appropriately selected statistical techniques in contexts that address important questions in interpretable ways. In fact, we will see that the data and statistical issues facing both of our subjects are far more similar than they are different, although their contents greatly differ. The process of doing statistics in research projects changes very little across a wide variety of questions and fields of interest.

The world of statistics starts with a question, not with data. The quality of the question in combination with the quality of the data greatly determines the quality of the results. No surprise there, but quality research questions can be surprisingly rare in otherwise well-conducted studies, which is a situation that often leads to ambiguity in the results.