Here are the expected and observed numbers for the various e+e and e+mu dilepton bins.  The background estimate is too high in many of the cases, but the same-sign same-flavor categories are reasonably close.  There are some problems with the code that stacks all the background contributions together, and I believe that that is where the differences are coming from.

DILEP_TCE_eP_eP nbg=550.536 data=469
DILEP_TCE_eP_eN nbg=12044.3 data=439
DILEP_TCE_eN_eN nbg=583.46 data=428
DILEP_CMUP_eP_eP nbg=331.307 data=302
DILEP_CMUP_eP_eN nbg=6261.74 data=284
DILEP_CMUP_eN_eN nbg=342.74 data=315
DILEP_CMX_eP_eP nbg=189.794 data=170
DILEP_CMX_eP_eN nbg=3783.66 data=173
DILEP_CMX_eN_eN nbg=185.485 data=201
DILEP_TCE_eP_mP nbg=65.4135 data=19
DILEP_TCE_eP_mN nbg=63.8633 data=26
DILEP_TCE_eN_mP nbg=58.6098 data=36
DILEP_TCE_eN_mN nbg=64.9233 data=30
DILEP_CMUP_eP_mP nbg=32.5401 data=12
DILEP_CMUP_eP_mN nbg=30.2969 data=24
DILEP_CMUP_eN_mP nbg=31.4779 data=16
DILEP_CMUP_eN_mN nbg=32.2473 data=14
DILEP_CMX_eP_mP nbg=16.3119 data=14
DILEP_CMX_eP_mN nbg=20.3123 data=8
DILEP_CMX_eN_mP nbg=15.8709 data=12
DILEP_CMX_eN_mN nbg=22.3682 data=13

-Scott