Three-item analysis: Hierarchical representation and treatment of missing and inapplicable data

Matrix-based methods, including parsimony programs, represent and treat missing data in the same way, such that every character-state is possible, and inapplicable data, such that every character state is impossible. This is because the hierarchical nature of homology assessments cannot be represented in taxon–character matrices. We show that the hierarchical representation of hypotheses of homology used in three-item analysis permits the accurate treatment of missing and inapplicable data.