Table 3. Generalised Linear Models (binomial errors) where the response variable that was modelled was the probability of a leopard attack on livestock.

The data set consists of both compensation claimants as well as random households.

Model	Deviance	AICc	ΔAICc	w
Dogs present+livestock protected+dogs present: livestock protected	119.76	128.03	0.00	0.27
Dogs present + livestock protected	123.04	129.21	1.17	0.15
Dogs present + number of goats + livestock protected + dogs present: livestock protected	119.01	129.43	1.39	0.14
Dogs present + number of other livestock + number of goats + livestock protected + dogs present: livestock protected	116.96	129.55	1.52	0.13
Dogs present + number of other livestock + livestock protected	121.88	130.16	2.12	0.09
Dogs present + number of goats + livestock protected	122.07	130.35	2.31	0.09
Dogs present + number of other livestock + number of goats + livestock protected	120.64	131.06	3.03	0.06
Livestock protected	128.06	132.14	4.11	0.03
Number of goats + livestock protected	127.23	133.40	5.37	0.02
Number of other livestock + livestock protected	128.03	134.19	6.16	0.01
Number of other livestock + number of goats + livestock protected	127.19	135.47	7.43	0.01
Dogs present + number of goats	191.08	197.25	69.21	0.00
Dogs present + number of other livestock + number of goats	190.84	199.12	71.08	0.00
Dogs present	195.85	199.93	71.90	0.00
Dogs present + number of other livestock	195.73	201.89	73.86	0.00
Number of goats	199.29	203.37	75.33	0.00
Number of goats + number of other livestock	198.98	205.15	77.11	0.00
Number of other livestock	203.15	207.24	79.20	0.00