Mean separation testing in variety trials when modeling assumptions are violated

Mentors:
Lead: Emily Griffith (Statistics, NC State) Collaborator: Michael Parker (Horticulture Science, NC State)

Outline:  
Agricultural variety trials help farmers select between different cultivars of crops to maximize crop yield and plant survival. These trials are increasingly important as factors like climate change and pesticide resistance may impact yield from older cultivars. Many of these trials are randomized complete block designs; analyzed using ANOVA followed by mean separation tests to rank the studied cultivars in order of increasing yield or other outcomes of interest. However, the data from variety trials may have missing observations (due to plant loss), heterogeneous variance (due to biological factors), and be lognormally distributed. Building on previous work that studied the impact of lognormally vs. normally distributed data on inferences drawn from mean separation tests, we will use simulation studies to explore the impact of missingness, unbalanced blocks, and heterogeneous variances on mean separation patterns. This work will use real datasets from horticulturalists as a foundation, and will have immediate impact on inferences drawn from variety trials.

Objectives:
(i) Investigate different forms of assumption violations based on real data
(ii) Design and run a simulation study to test the impact of specific violations identified in (i).
(iii) Describe and justify the results found in the simulation study.

Outcomes:
The students will present a poster summarizing their research and submit a paper for publication in a peer-reviewed journal.

References:
1. Agbangba, C.E., Sacla Aide, E., Honfo, H., Glèlè Kakai, R. (2024). On the use of post-hoc tests in
environmental and biological sciences: A critical review. Heliyon 10(3):e25131.
2.  Montgomery, D. C. (2017). Design and analysis of experiments. John Wiley & Sons, Inc. (Chapters 2
and 15, specifically)

3.  Parker, M.L, Hoyt, T., and Clark, B. (2014). Evaluating apple replant strategies in the southeastern United States. Acta Hortic. 1058, 645-650.

4.  Petrinovich L.F., Hardyck C.D. (1969). Error rates for multiple comparison methods: some
evidence concerning the frequency. Psychological Bulletin, 71(1), 43–54.