COEFFICIENTS OF ASSOCIATION AND SIMILARITY, BASED ON BINARY (PRESENCE‐ABSENCE) DATA: AN EVALUATION

• DOI: 10.1111/j.1469-185x.1982.tb00376.x
• Publication Date: 2008-01-22T03:44:25Z
• Authors: ZDENEK HUBÁLEK
• Keywords: Jaccard index; Similarity (geometry)

Summary Forty‐three association (similarity) coefficients were collected and evaluated in this survey. Some of them are synonyms or direct correlates with earlier described indices (A 8 , A 9 12 31 33 ), others mere transforms from one range values to another 10 24 ). Several incompatible suggested admissibility conditions the minimum‐maximum value 13 16 27 28 29 symmetry 1 2 26 discrimination between positive negative ) monotonicity (χ 19 ); 17 yields very low erratic values. As a result, 23 excluded remaining 20 measures subjected an empirical trial on interspecific data among fungi genus Chaetomium use cluster analysis. The classification produced five main clusters related coefficients, several subgroups. It was then demonstrated that representative different yield dendrograms 34 14 possibly also 36 40 seemed be less sensible. set generally work well (at least association) comprises 4 (Jaccard), (Dice‐Sφrensen), 7 (Kulczyński), 11 (Driver‐Kroeber‐Ochiai) and, some reservation 30 (Pearson tetrachoric) 32 (Baroni‐Urbani‐Buser). For purposes, however, other ‘admissible’ would more optimal, choice measure should nature data. is tentatively three so alternative used results compared same basis; moreover, significance tests carried out whenever possible.