A glimpse of GLMMIX (with a peek at R) for use in horticultural research

Statistics, like horticulture, is a science that is evolving.
The purpose of this workshop was to introduce the latest methodology for design and analysis of blocked experiments.
ANOVA, when applied to non-normal NDASH more properly called non-Gaussian NDASH data common in horticultural research, is not recommended as it produces inaccurate mean estimates and unreliable hypothesis test results.
Generalized linear mixed models (GLMM), implemented by the SAS^® procedure GLIMMIX, or its R equivalent, lme4, allow the researcher to identify the appropriate distribution and analyze the data with more accuracy.
Using examples of block designs for a greenhouse situation in which strawberries in pots are grown on benches, we showed how to use GLMM methods to optimize the design.
We also illustrated the differences between ANOVA- and GLMM-based analysis, and demonstrated how to select and interpret appropriate analyses.
Emphasis was placed on engaging a statistician early and often during the research process.

XXIX International Horticultural Congress on Horticulture: Sustaining Lives, Livelihoods and Landscapes (IHC2014): Plenary Sessions of IHC2014 and VII International Symposium on Education, Research Training and Consultancy

E.T. Paparozzi, W.W. Stroup

statistics, GLMM, ANOVA, strawberries, experiment design, analysis, repeated measures

https://doi.org/10.17660/ActaHortic.2016.1126.32