Studying the correlation of type Ia supernova rates (SNR) with host galaxy properties is an important step in understanding the exact nature of type Ia supernovae. We use SNe Ia from the SDSS-II sample, spectroscopically determined masses and star formation rates, and a new maximum likelihood method, to fit the Scannapieco and Bildsten rate model $SNR = A{\times}M + B{\times}SFR$, where $M$ is galaxy mass and $SFR$ is star formation rate. We find $A = 3.5^{+0.9}_{-0.7}\times10^{-14}(\text{SNe/yr})(M_{\odot})^{-1}$ and $B =1.3^{+0.4}_{-0.3}\times10^{-3}(\text{SNe/yr})(M_{\odot}$yr$^{-1})^{-1}$, assuming overall efficiency of 0.5. This is in reasonable agreement with other determinations. However we find strong evidence that this model is a poor fit to other projections of the data: it fails to correctly predict the distribution of supernovae with host mass or SFR. An additional model parameter is required; most likely this parameter is related to host galaxy mass. Some implications of this result are discussed.