IMPACTO DA VIOLAÇÃO DOS PRESSUPOSTOS DA ESTATÍSTICA PARAMÉTRICA EM ESTUDOS MORFOMÉTRICOS

Impact of Assumption Violations in Parametric Statistics in Morphometric Studies

Abstract

The objective of this study was to demonstrate the importance of characterizing data according to their probability distribution and the consequent impact on parametric test results when the data distribution model is poorly specified. The dataset comprises abdominal length measurements of lobsters categorized by qualitative variables: sex and month. Data analysis proceeded through five steps: (1) comparison of means without prior characterization of the continuous random variable; (2) characterization of the continuous random variable by probability distribution; (3) data transformation and recharacterization of the continuous random variable; (4) application of a generalized linear model to compare factors; and (5) assessment of the impact of violating parametric statistical assumptions on factor comparisons. Results analysis revealed that violating parametric statistical assumptions, such as those underlying Student's t-test commonly used in biological research, can lead to erroneous conclusions. Hence, correctly applying appropriate statistical tools is crucial for ensuring result reliability and accurate inference.

Keywords: Student’s t-test, Normality, Variance, Generalized linear model.