We measure the two–point redshift–space correlation function of loose groups of galaxies , \xi _ { GG } ( s ) , for the combined CfA2 and SSRS2 surveys .
Our combined group catalog constitutes the largest homogeneous sample available ( 885 groups ) .
We compare \xi _ { GG } ( s ) with the correlation function of galaxies , \xi _ { gg } ( s ) , in the same volume .
We find that groups are significantly more clustered than galaxies : < \xi _ { GG } / \xi _ { gg } > = 1.64 \pm 0.16 .
A similar result holds when we analyze a volume–limited sample ( distance limit 78 h ^ { -1 } Mpc ) of 139 groups .
For these groups , with median velocity dispersion \sigma _ { v } \sim 200 km s ^ { -1 } and mean group separation d \sim 16 h ^ { -1 } Mpc , we find that the correlation length is s _ { 0 } = 8 \pm 1 h ^ { -1 } Mpc , which is significantly smaller than that found for rich clusters .
We conclude that clustering properties of loose groups of galaxies are intermediate between galaxies and rich clusters .
Moreover , we find evidence that group clustering depends on physical properties of groups : correlation strengthens for increasing \sigma _ { v } .