Advanced statistical analysis of small-plot canola variety trial data

The Challenge

Current analysis of small plot trials use simple statistics to express variety performance as a percentage of designated check variety value at each site-year, and over all site years. Problems arise from this approach in that:

• Comparisons among varieties other than the check(s) are not necessarily valid
• There may be problems choosing the check(s)
• It requires consistent performance of the check(s) over locations and years

The Project

Use mixed model statistical analysis (MMSA) to avoid problems and gain more information from crop variety trial data. MMSA will provide more accurate model-based estimate of the variety mean yield and is better able to accommodate highly unbalanced datasets.

The Results

Mathematical and statistical theory indicate that MMSA will always provide better or equal results to an arithmetic mean-based approach. When data is limited, and/or one out of ten years was unusual, MMSA proves to be more accurate and responsive over arithmetic means.

Grower Benefits

When choosing a canola variety, data presented will be more concise, relatable and useful, which will provide growers with better insight as to which variety is best for their operation.