Proficiency test is a type of inter-laboratory program that allows determining the performance of a laboratory by comparing it to at least other two laboratories, through measurements in homogeneous or similar materials under predetermined conditions. In the area of seeds, for the process of maintaining accreditation, the International Seed Testing Association (ISTA) requires the participation of certified seed laboratories in proficiency tests. The statistical technique used by ISTA for the assessment of proficiency test results is the Z-score method (ISTA, 2007). The research problem approached in this paper was to investigate and assess statistical technique of Cluster Analysis, besides the Z-score method, that would prove adequate to analyze the results of proficiency tests in the area of seeds. In order to generate the database, three simulations have been carried out in software R, each representing the results of germination test in oil radish seeds, with twelve participating laboratories. The twelve laboratories were arranged in two discrepant groups regarding some statistical parameter(s) such as variance and covariance average and matrix. In the case of Z-score method, precision and accuracy were calculated for each laboratory and simulation, and rating and grade were attributed. In the Cluster Analysis was used squared Euclidean distance and average method, then dendrograms were built and cophenetic correlations were calculated for each simulation. The technique of Cluster Analysis, in general, managed to distinguish the two groups, presenting themselves as potential procedures to be used for proficiency tests in the area of seeds. The Z-score method did not manage to adequately separate the groups, and further studies become necessary to assess the performance of that procedure in the simulated situations.
References:
International Seed Testing Association. The ISTA Proficiency Test Programme. Switzerland, 2007.
Keywords: Proficiency test; Cluster Analysis; Z-score method; Oil radish seeds
Biography: Master an PhD in Statistics and Agropecuary Experimentation at UFLA-MG. 21 year as an in service teacher. Educational researcher and materials developer in Statistics Education using software R. At present day, teacher and researcher in post-graduated in Mathematics Education at the Universidade Bandeirante de São Paulo.