Source-coded data partitioning of a compressed video stream protects more important data, reducing the risk of error corruption in wireless networks. However, because the data partitions of a video slice are each assigned to different network packets, in congestion-prone networks the increased number of packets per slice and their size disparity may increase the packet loss rate from buffer overflows. This paper recommends packet-size dependent scheduling as a relatively simple way of alleviating the problem, which problem remains real whenever there is an adverse distribution of packet sizes entering an access network. The paper also contributes an analysis of partition and packet sizes as a prelude to scheduling regimes.