If you're making your deciding factor a size of 3, then you're eliminating some of the planetary bodies in this system. (I know, a sample of 1 isn't a sample.)
I can agree with probability marking whether a body is a planet or a moon. But if a system has a sufficently large and violent parent star, the probability of the mainworld being a moon should increase. My thought being that a system with a big star would put out more hard radiation than a smaller one (as a rule), thereby requiring habitable worlds to live within the magnetosphere of a larger world. Smaller worlds would lack the protective magnetic field that a larger world possesses, and with increased solar output, these smaller worlds would be ravaged by the hard radiation of the parent star.
BUT, with all the new observations coming in, some of thses guesses and ideas could be proven wrong tomorrow.