For a given channel, with certain characteristics, and for a certain value of data rate, the outage probability should be higher if frame length is smaller, particularly from outages due to fadings (when signal undergoes deep fades).
Larger frames does not fit into real time applications such as voice and video. Till frame is filled, data is not transmitted. Therefore, the larger the frame size, the more delay. Voice transmission quality suffers with delay. Same is true with Video.
Therefore, the frame size is decided based upon the applications.
If latency is not an issue, theoretically frame size can be large.
Throughput with error free data transmission is the goal in wireless communication. Bursty errors are inherent in wireless communication. Therefore, the frame size should consider these errors for achieving best possible throughput.
The question is too ambiguous. Outage rate is more related with the link not the packets. Packets lost caused by fast fading can be alleviated by Error forward coding, i.e. to enhance the link quality by sacrificing bit rate. But outage caused by shadowing and pathloss can not be avoided by modifying frame length.