We combine the CfA3 supernova Type Ia ( SN Ia ) sample with samples from the literature to calculate improved constraints on the dark energy equation of state parameter , w . The CfA3 sample is added to the Union set of Kowalski et al . ( 33 ) to form the Constitution set and , combined with a BAO prior , produces 1 + w = 0.013 ^ { +0.066 } _ { -0.068 } ( 0.11 ~ { } syst ) , consistent with the cosmological constant . The CfA3 addition makes the cosmologically-useful sample of nearby SN Ia between 2.6 and 2.9 times larger than before , reducing the statistical uncertainty to the point where systematics play the largest role . We use four light curve fitters to test for systematic differences : SALT , SALT2 , MLCS2k2 ( R _ { V } = 3.1 ) , and MLCS2k2 ( R _ { V } = 1.7 ) . SALT produces high-redshift Hubble residuals with systematic trends versus color and larger scatter than MLCS2k2 . MLCS2k2 overestimates the intrinsic luminosity of SN Ia with 0.7 < \Delta < 1.2 . MLCS2k2 with R _ { V } = 3.1 overestimates host-galaxy extinction while R _ { V } \approx 1.7 does not . Our investigation is consistent with no Hubble bubble . We also find that , after light-curve correction , SN Ia in Scd/Sd/Irr hosts are intrinsically fainter than those in E/S0 hosts by 2 \sigma , suggesting that they may come from different populations . We also find that SN Ia in Scd/Sd/Irr hosts have low scatter ( 0.1 mag ) and reddening . Current systematic errors can be reduced by improving SN Ia photometric accuracy , by including the CfA3 sample to retrain light-curve fitters , by combining optical SN Ia photometry with near-infrared photometry to understand host-galaxy extinction , and by determining if different environments give rise to different intrinsic SN Ia luminosity after correction for light-curve shape and color .