Hermits show different preferences depending on species. For example, I've seen a strong preference in Calcinus seurati for very globose Nerite and Turbinid and shells over all else, while the small Clibanarius species seem to show a preference for more conical Cerithid shells. They will accept globose shells too with little issue if they need a size upgrade, but given two of the same size strangely seem to prefer the clunkier, conical one. The angle of the aperature is also a factor. Both the Calcinus and small Clibanarius species I have kept consistently rejected Astraea/Lithopoma shells when other similarly-sized opportunities existed. Lumpy, ridged, or "toothed" apertures also seem to be undesirable for many species.
As fishy mentioned, Ciliopagurus strigatus needs long, narrow apertures characteristic of cone snail species (and some conches). Paguristes puncticeps is another, although uncommon species in the trade that is also flattened and seems to prefer elongate apertures. Most of the commonly available species need rounded apertures to avoid getting stuck while still having good mobility and protection. If the apertures are big enough in terms of "height" but still too elongate side-to-side, hermits may have trouble moving the shell around and may also still be somewhat exposed from the side when they retract into the shell (easier to get legs pulled off by another crustacean).
As far as bigger being better...to date, Dardanus megistos is the only species I have kept that demonstrates that tendency, and only in a relatively calm environment with ample shells. They seem to like to put on the biggest shell possible for the purpose of doing pushups and other dominance displays at other hermits. However, if they feel threatened (such as when a larger crab answers the display by charging over), they waste no time jumping out of the big shell and into something smaller that they can run around with more easily. In other species, bigger is usually only better to the extent that it still allows both growth and reasonable mobility.