The objectives of this thesis are to present novel approaches for optimising the construction of multi environment trial (MET) datasets from a series of plant variety trials. These include evaluating varieties in designed trials at various locations and typically across many years. The MET datasets are then analysed to evaluate how well each variety performs in each environment. Although sophisticated and relevant statistical analyses have been proven to increase the reliability of predicted variety by environment (VE) effects, there has been little research into how to construct an appropriate dataset. This thesis fills a void in the literature by providing information-based diagnostics for the optimal construction of the MET dataset. The approaches are demonstrated using two motivating datasets: the first is a Oat (Avena sativa) dataset and the other is a Durum wheat (Triticum durum L. ssp. Durum Desf.) dataset. The former is used as an example of a dataset with independent variety effects, whereas the latter is used as an example of a dataset with related variety effects. These are also used for their attributes in the development of real-world grounded simulation studies to examine the performance of the proposed diagnostics and also to investigate established methodologies and concerns.