Table 2.

Analyses summary

	Tracking data used	Tests	Response variable	Fixed effect	Random effect	Sample size
Association of last week with grouse moors	n = 24,447 N = 10,974, fixes	Binomial GLMM	Binary: terminal week (1) or an alive week (0)	—Weekly proportion of fixes on grouse moors —Days since tagging (max. per week)	Bird ID	n = 1475, N = 692 weeks
Landscape-scale analysis	n = 23,357, N = 10,800, fixes on UK mainland and Isle of Man	Binomial GLM	Two vector cbind: number of fixes in terminal week, number of other fixes, per 20 km2 grid cell	Proportion of 1 km2 cells with grouse moor cover in 20 km2 grid	na	n = 309, N = 244 grid squares
Protected area analysis	n = 5960 N = 4940, fixes in protected areas in England	Binomial GLM	Two vector cbind: number of fixes in terminal week, number of other fixes, per PA	Proportion of 1 km2 cells with grouse moor cover inside PAs	na	n/N = 8 protected areas

Three main analyses used to explore whether death or disappearance of satellite-tracked hen harriers were associated with grouse moor habitat. Sample size of fixes used per analysis are described, where (n) includes fixes from all harriers and (N) is the sample size for repeat analyses using only suspected (SIM) and known (I) illegal mortalities