From work done with sybr and taqman assay design as well as highly multiplex PCR for NGS, we know there are several factors that bias against an amplicon.
Size (larger ones are biased against in a multiplex reaction
GC content of amplicon - higher than ~60% is a frequent cause of bias or failure
GC content and distribution of GC in the primers - proper design should take care of that
Secondary structure of primers or the single stranded amplicon.
Stressing the multiplex reaction too much (insufficient taq, Mg, dNTP, too much primer, too little primer, etc).