Smith, MSullivan, MWiseman, PKessler, RScolnic, DBrout, DD'Andrea, CBDavis, TMFoley, RJFrohmaier, CGalbany, LGupta, RRGutiérrez, CPHinton, SRKelsey, LLidman, CMacaulay, EMöller, ANichol, RCNugent, PPalmese, APursiainen, MSako, MSwann, EThomas, RCTucker, BEVincenzi, MCarollo, DLewis, GFSommer, NEAbbott, TMCAguena, MAllam, SAvila, SBertin, EBhargava, SBrooks, DBuckley-Geer, EBurke, DLRosell, ACKind, MCCostanzi, Mda Costa, LNde Vicente, JDesai, SDiehl, HTDoel, PEifler, TFEverett, SFlaugher, BFosalba, PFrieman, JGarcía-Bellido, JGaztanaga, EGlazebrook, KGruen, DGruendl, RAGschwend, JGutierrez, GHartley, WGHollowood, DLHonscheid, KJames, DJKrause, EKuehn, KKuropatkin, NLima, MMacCrann, NMaia, MAGMarshall, JLMartini, PMelchior, PMenanteau, FMiquel, RPaz-Chinchón, FPlazas, AARomer, AKRoodman, ARykoff, ESSanchez, EScarpine, VSchubnell, MSerrano, SSevilla-Noarbe, ISuchyta, ESwanson, MECTarle, GThomas, DTucker, DLVarga, TNWalker, AR2020-04-232020-04-23https://hdl.handle.net/10161/20491We present improved photometric measurements for the host galaxies of 206 spectroscopically confirmed type Ia supernovae discovered by the Dark Energy Survey Supernova Program (DES-SN) and used in the first DES-SN cosmological analysis. Fitting spectral energy distributions to the $griz$ photometric measurements of the DES-SN host galaxies, we derive stellar masses and star-formation rates. For the DES-SN sample, when considering a 5D ($z$, $x_1$, $c$, $\alpha$, $\beta$) bias correction, we find evidence of a Hubble residual `mass step', where SNe Ia in high mass galaxies ($>10^{10} \textrm{M}_{\odot}$) are intrinsically more luminous (after correction) than their low mass counterparts by $\gamma=0.040\pm0.019$mag. This value is larger by $0.031$mag than the value found in the first DES-SN cosmological analysis. This difference is due to a combination of updated photometric measurements and improved star formation histories and is not from host-galaxy misidentification. When using a 1D (redshift-only) bias correction the inferred mass step is larger, with $\gamma=0.066\pm0.020$mag. The 1D-5D $\gamma$ difference for DES-SN is $0.026\pm0.009$mag. We show that this difference is due to a strong correlation between host galaxy stellar mass and the $x_1$ component of the 5D distance-bias correction. To better understand this effect, we include an intrinsic correlation between light-curve width and stellar mass in simulated SN Ia samples. We show that a 5D fit recovers $\gamma$ with $-9$mmag bias compared to a $+2$mmag bias for a 1D fit. This difference can explain part of the discrepancy seen in the data. Improvements in modeling correlations between galaxy properties and SN is necessary to determine the implications for $\gamma$ and ensure unbiased precision estimates of the dark energy equation-of-state as we enter the era of LSST.astro-ph.COastro-ph.COastro-ph.GAJournal article2020-04-23