Table 1.

Meta-regression of potential moderators of housing effects on stress-sensitive disease

	Test statistic	p
Species	F1,154 = 0.2222	0.6381
Sex	F2,154 = 0.0160	0.9841
Social status	F2,154 = 0.6564	0.5201
Disease	F4,154 = 1.3522	0.2532
Measure: infarct volume versus others	F1,154 = 22.7385	<  0.0001
Species × sex	F2,154 = 1.6491	0.1956
Species × social status	F2,154 = 0.5794	0.5615
Sex × social status	F4,154 = 0.6208	0.6484
Resource category	F3,139 = 0.8280	0.4806
Resource category × species	F3,139 = 1.0409	0.3766
After removal of ‘red flags’
Species	F1,91 = 0.0351	0.8517
Sex	F2,91 = 0.2542	0.7761
Social status	F2,91 = 0.4339	0.6493
Disease	F4,91 = 2.5952	0.0415
Measure: infarct volume versus others	F1,91 = 15.8439	0.0001
Species × sex	F2,91 = 0.9884	0.3761
Species × social status	F2,91 = 0.9929	0.3745
Sex × social status	F4,91 = 0.2229	0.9250
Resource category	F3,83 = 2.5128	0.0641
Resource category × species	F3,83 = 0.8890	0.4504

Results from a random-effects meta-regression investigating potential moderators of housing effects (effects of conventional housing versus housing ‘enriched’ with resources supporting species-typical behaviour) on stress-sensitive disease (standardized mean differences). (See Additional file 10 for a replicate excluding study weights). Bold p values are significant at p < 0.05.