Multivariate analysis of variance: An advanced chemometric approach to differentiate dose‐dependent antioxidant activities of grape ( Vitis labruscana ) skin extracts

Abstract

Multivariate analysis of variance (MANOVA) was performed to differentiate dose-dependent antioxidant properties of grape skin extracts and compare the detailed analysis of MANOVA using 12 multivariate post hoc tests. The antioxidant activities were assessed by in vitro antioxidant activity assays, and data were processed using MANOVA statistical and post hoc tests. Wilks's Λ reported significant differences, F(20, 20.85) = 46.57 (p < .0001, Wilk's Λ = 0.03 × 10−3), between dose-dependent and radical scavenging activities of grape skin extracts. The results of post hoc analyses demonstrated that Tukey's (single-step) and Duncan's (step-down) multiple range tests were suitable methods for evaluating multiple mean comparisons. Tukey's and Duncan's multiple range methods were recommended for MANOVA post hoc analysis. The new findings revealed the use of statistical analysis in order to prove scientific hypotheses. Moreover, MANOVA statistical analysis is an alternative for choosing MANOVA for comparison of multiple variables. Practical applications The study was designed to differentiate dose-dependent antioxidant properties of grape skin extracts and compare the detailed analysis of MANOVA using 12 multivariate post hoc tests. MANOVA statistical tests and post hoc tests could be an alternate and easy statistical approach for evaluating the mean differences among the samples. Moreover, it is also provided a theoretical basis for the selection of best statistical methods in order to prove research hypotheses.