Jacob CohenStatistical Power Analysis for the Behavioral Sciences

This is the bible for research in the behavioral sciences -- THE place to start learning. Superb explanation of effect size - how and why to calculate it and how to deal with unexpected problems with the data. Also, I return to it often as a reference. It does NOT have the latest information on new approaches and methods that have been developed in the last 10 - 15 years. But I go back to it to get oriented when I get stuck in trying to learn a new method (something I read in a journal or something I encounter in software packages). Before I studied this book (first ed., early 1980's) I thought that statistical significance was all that mattered for successful research studies. This opened my eyes to the importance of effect size. NOTE -- this error of omission (reporting significance but not effect size) is still being made in today's journals(!!!), maybe a little less frequently. People in a hurry still look mostly at the mean value of data sets, not the range (or other aspects) of the distribution.

Un ouvrage qui a pris de l'age (beaucoup!!) mais qui reste efficace pour ceux confrontés aux calculs de puissanceLes calculs y sont abordés sous différents aspects (comparaisons de moyennes, de proportion, de tableau de contingence, application a la régression) avec le minimum théorique nécessaire.Beaucoup de tables obsolètes au regard de l'offre logiciel disponible mais le raisonnement du calcul de puissance est bien décrit

This is an amazing book that is not appreciated enough. Finally broke down and actually spent the money to buy it because it is so valuable. Every student, professor or researcher doing any sort of research that uses statistics. Absolutely critical to understanding and interpreting modern statistics. I was able to read through this book and understand what Cohen was talking about after an undergraduate statistics course. Really simple to understand and really able to see the implications of the effect size. Also, you realize that the p-value is a really meaningless score! Best of luck to all!

The Bible of statistical power and effect sizes.

I did find more straight forward means of calculating effect size and sample size at a free stats web site. So, purchase of this text was a bit of a waste.