Table 1.

Contributions of each moderating variable to account for variance in effect sizes across studies

Moderator	F	Df1, Df2	p
Population
Animal species	2.613	(10, 145)	<.0001***
Animal class	5.811	(3, 152)	.0009***
Human vs. Non‐human	7.555	(2, 153)	.0007***
Training and testing
Log training length	12.149	(1, 154)	<.0001***
Stimulus modality	0.095	(2, 153)	.909
Test response	1.624	(10, 145)	.105
Test type	3.698	(1, 154)	.056
Surface‐level properties
Categories in language	0.0001	(1, 154)	.992
Number of unique vocabulary items	3.021	(1, 154)	.084
Structural properties
Repetition of items	14.162	(1, 154)	.0002**
Adjacent dependencies	0.238	(1, 154)	.627
Non‐adjacent dependencies	0.118	(1, 154)	.608

F is the statistic for testing whether the moderator accounts for some heterogeneity between studies; p is the significance for the F‐test ***p < .001, **p < .01, *p < .05. Note that Animal Class distinguishes birds, non‐human mammals, human adult, and human child. Animal species also distinguishes human adult and human child.