An earlier paper by Bell et al. [2008] showed satellite evidence that average summertime (1998-2005) rainfall over the non-coastal southeast U.S. varied with the day of the week in a statistically significant way, with the maximum occurring midweek (Tue-Thu). An explanation was proposed in which the recurring midweek increase in air pollution over the area causes a shift in the drop-size distribution in clouds to smaller sizes as the clouds develop. The smaller droplets could be carried to higher altitudes where their freezing releases additional latent heat, invigorating the storms. Evidence for this phenomenon was provided by storm-height distributions obtained from the Tropical Rainfall Measuring Mission radar, but the statistical significance of the midweek increase in storm heights was unclear. An improved statistical analysis of the storm-height distributions is provided here, indicating that the probability that storms climb above altitudes 7-15 km is increased midweek relative to weekends (Sat-Mon) for afternoon storms (1200-2400 local time). The morning storm heights, on the other hand, are found not to exhibit statistically significant shifts, which would be consistent with the above explanation. Morning storm statistics are also found to be much more sensitive than afternoon storm statistics to the exact area over which the averages are taken.