Improved two-step analysis of germination data from complex experimental designs. Issue 3 (September 2020)

• Record Type: Journal Article
• Authors: Jensen, Signe M.; Wolkis, Dustin; Keshtkar, Eshagh; Streibig, Jens C.; Ritz, Christian
• Is Part Of: Seed science research. Volume 30:Issue 3(2020)
• Page Start: 194
• Page End: 198
• Publication Date: 2020-09
• Subjects: factorial designs, -- germination analysis, -- log-logistic models, -- replication, -- simultaneous modelling, -- visualization; Seeds -- Periodicals; 631.521
• DOI: 10.1017/S0960258520000331
• Languages: English
• ISSNs: 0960-2585

Abstract

Abstract: Germination experiments are becoming increasingly complex and they are now routinely involving several experimental factors. Recently, a two-step approach utilizing meta-analysis methodology has been proposed for the estimation of hierarchical models suitable for describing data from such complex experiments. Step 1 involves fitting models to data from each sub-experiment, whereas Step 2 involves combination estimates from all model fits obtained in Step 1. However, one shortcoming of this approach was that visualization of resulting fitted germination curves was difficult. Here, we describe in detail an improved two-step analysis that allows visualization of cumulated data together with fitted curves and confidence bands. Also, we demonstrate in detail, through two examples, how to carry out the statistical analysis in practice.