103. Mićić N., Kurtović M., Knezović Z., Bosančić B.: Cilj istraživanja i logičko-matematička argumentacija rezultata biometričkih analiza. Poljoprivredno-prehrambenog fakulteta, Univerziteta u Sarajevu, God. LIX, broj 64/2
The first step in defining the experimental approach in biometrical analysis is defining the realistic and expected treatment contribution [(x̄ treatment - x̄ control), i.e. Δx̄ T1- T2] which is assessed as being functional and indicative regardless of the statistical significance of the established differences. Hence, when the experimental results confirm the indicative treatment effects it is expected that those are also confirmed by the appropriate statistical tests. The same is expected also in case when this relationship is inversed. However, in the case when applied statistical test documents conclusions which are contrary to the indicative treatment contribution, then the experimental results and the defined research goals are in an absurd relation and have to be additionally analysed. The first step in the analysis of these relations is checking the variability of the observed experimental units within the samples, i.e. verification of the representability of the experimental samples by the analysis of allowed variations in the argumentation of central tendencies of the treatments. Ignoring the absurd relations regarding the functional or indicative treatment effect as the set research objective and the statistical significance of the differences supporting the research objective in conclusion can make the research itself absurd.