What works for one may not always work for another. Even slight differences in habitat, food, stresses, and living styles can change the need for dimorphism. The ant genus pheidole has a highly dimorphic caste because minor sisters are small and cannot carry out larger tasks. So more expensive, and larger, sisters are made to help with large tasks and for defending the nest from army ant intruders that seem to favor running raids on pheidole colonies. A genus such as pogonomyrmex lack any serious polymorphism because there really is no need for it yet the two listed above live right next to each other at times. I have not studied mantids as much as I should but there is going to be a specific reason, a very specific reason, why orchids exhibit huge dimorphism that other mantids lack.
Every insect out there is only after one thing, to reproduce and more often than not, they will have very odd and clever ways of being successful. That's why you can't have a general theory that matches for all species of every mantid genus on this planet because you can't expect species from Africa to live the same life as species from Australia or China. Every evolutionary change is to benefit the population because it works, they don't choose to change, it's just smaller males seem to work well for them, so they've become the norm of the species.