Considerations for missing data, outliers and transformations in permutation testing for ANOVA, ASCA(+) and related factorizations
Authors:
Oliver Polushkina Merchanskaya,
Michael D. Sorochan Armstrong,
Carolina Gómez Llorente,
Patricia Ferrer,
Sergi Fernandez-Gonzalez,
Miriam Perez-Cruz,
María Dolores Gómez-Roig,
José Camacho
Abstract:
Multifactorial experimental designs allow us to assess the contribution of several factors, and potentially their interactions, to one or several responses of interests. Following the principles of the partition of the variance advocated by Sir R.A. Fisher, the experimental responses are factored into the quantitative contribution of main factors and interactions. A popular approach to perform thi…
▽ More
Multifactorial experimental designs allow us to assess the contribution of several factors, and potentially their interactions, to one or several responses of interests. Following the principles of the partition of the variance advocated by Sir R.A. Fisher, the experimental responses are factored into the quantitative contribution of main factors and interactions. A popular approach to perform this factorization in both ANOVA and ASCA(+) is through General Linear Models. Subsequently, different inferential approaches can be used to identify whether the contributions are statistically significant or not. Unfortunately, the performance of inferential approaches in terms of Type I and Type II errors can be heavily affected by missing data, outliers and/or the departure from normality of the distribution of the responses, which are commonplace problems in modern analytical experiments. In this paper, we study these problem and suggest good practices of application.
△ Less
Submitted 13 August, 2024;
originally announced August 2024.
