Analysis of Generalized Linear Mixed Models in the Agricultural and Natural Resources Sciences is a new world-wide release at an important time of change in the research community. It demonstrates, through examples, the design and analysis of mixed models for non-normally distributed data and challenges traditional statistical methodology. It is written by a team of authors who are part of a multi-state project to educate scientists in the agricultural and natural resources sciences about modern statistical methodology. One of its lead writers, Edward E. Gbur says, "There's a gap between statistical theory and practice and the statistical methodologies currently being used within the agricultural and natural resources communities. There needs to be a change in the standard statistical operating procedure from the last decade."
Excerpts below from Analysis of Generalized Linear Mixed Models in the Agricultural and Natural Resources Sciences by Edward E. Gbur, Walter W. Stroup, Kevin S. McCarter, Susan Durham, Linda J. Young, Mary Christman, Mark West and Matthew Kramer as published by the American Society of Agronomy, Soil Science Society of America and Crop Science Society of America. It is available now at www.SocietyStore.org.
Traditional statistical methods have been developed primarily for normally distributed data. Generalized linear mixed models extend normal theory linear mixed models to include a broad class of distributions, including those commonly used for counts, proportions, and skewed distributions. With the advent of software for implementing generalized linear mixed models, we have found researchers increasingly interested in using these models, but it is easier said than done. Our goal is to help those who have worked with linear mixed models to begin moving toward generalized linear mixed models. The benefits and challenges are discussed from a practitioner's viewpoint. Although some readers will feel confident in fitting these models after having worked through the examples, most will probably use this book to become aware of the potential these models promise and then work with a professional statistician for full implementation, at least for their first few applications.
Efficiency is a particularly important issue now when public research universities and other research entities in the agricultural sciences face ongoing fiscal constraints, tight resources, and shrinking budgets that are unlikely to change in the foreseeable future. If generalized linear mixed model based methods can achieve higher quality information with the same amount of data or information of equal quality with less data as the examples demonstrate, then they can and should be used.
As research grows in complexity and the penalty becomes increasingly severe for the kinds of inaccuracy demonstrated in the examples, what passed for standard methodology a decade or two ago will become increasingly unacceptable.