Understanding The New Statistics Effect Sizes Confidence Intervals And Metaanalysis Geoff Cumming by Geoff Cumming 9780203807002, 9780415879675, 9780415879682, 0203807006, 0415879671, 041587968X instant download after payment.

The “new statistics” of this book’s title are not themselves new, but adopting them as the main way to analyze data would, for many students and researchers, be new. It would also be an excellent development. Let me explain.
“Children in the new program made significantly greater gains, p < .05.” In many disciplines, that’s the standard way to express a conclusion. It relies on null hypothesis significance testing (NHST), which uses p < .05 or p < .01 to establish a result as statistically significant. However, NHST is a very limited way to analyze and interpret data, because it focuses on the narrow question, “Is there an effect?” Other disciplines, including physics and chemistry, seldom use NHST, and usually report results as estimates by saying, for example: “the melting point of the plastic was 85.5 ± 0.2°C.” That’s usually much more informative than a statement that an effect is, or is not, statistically significant. The main message of this book is that we should shift emphasis as much as possible from NHST to estimation, based on effect sizes and confidence intervals. We should to some extent join physics, chemistry, and other disciplines that make enormous scientific progress with little use of NHST.