Speckle contrast imaging enables rapid mapping of relative blood flow distributions using camera detection of back-scattered laser light. However, speckle derived flow measures deviate from direct measurements of erythrocyte speeds by 47 $\pm$ 15% (n = 13 mice) in vessels of various calibers. Alternatively, deviations with estimates of volumetric flux are on average 91 $\pm$ 43%. We highlight and attempt to alleviate this discrepancy by accounting for the effects of multiple dynamic scattering with speckle imaging of microfluidic channels of varying sizes and then with red blood cell (RBC) tracking correlated speckle imaging of vascular flows in the cerebral cortex. By revisiting the governing dynamic light scattering models, we test the ability to predict the degree of multiple dynamic scattering across vessels in order to correct for the observed discrepancies between relative RBC speeds and multi-exposure speckle imaging estimates of inverse correlation times. The analysis reveals that traditional speckle contrast imagery of vascular flows is neither a measure of volumetric flux nor particle speed, but rather the product of speed and vessel diameter. The corrected speckle estimates of the relative RBC speeds have an average 10 $\pm$ 3% deviation in vivo with those obtained from RBC tracking.