The most common reason for this is that H264 encoded content was passed into Stomp.
H264 "loses" data as it encodes. The result is a series of images that are actually less well defined. Most of the time humans can't see it, but a computer sure can.
Stomp and other compressors like to see "original" video. The best compression is obtained by taking original frames and compressing them using Multipass H264 (the default H264 setting in Stomp is multipass, by the way).
So when you feed H264 video into Stomp, it has a really hard time trying to make it smaller. Each frame contains lots of visual artifacts from the original compression that don't re-compress well.
What's the solution?
For iShowU: Encode using ProRes or JPEG and then perform an H264 encode in Stomp. That'll yield the best (smallest) file size.
In general, the solution is to feed non-lossy (pure) footage into Stomp. That way the compressors will work far better.
Also, if you are using the standard Quicktime options, H.264 seems to work better (produces smaller files that look better) than MPEG4.