The primary difference between realism, constructive empiricism (CE), and anti-realism is where these approaches rest on the spectrum of ontology and explanation.  Realism takes theoretical commitments of science to be real, and not just [disguised] abbreviations for observational claims, or useful fictions we create to organize observations.[1]  Anti-realism is contrary to realism.  Instead of ‘X is an unobservable and X is real’, a la realism, anti-realism purports, ‘X is an unobservable and X is non-real.’  Both schools will recognize that, yes, X is an unobservable but they disagree on the ontic category.  The category of ontology becomes muddled, if not superfluous, when referring to unobservable entities.  An electron is a useful fiction.  Thus, whatever X, if X is commonly referred to what is considered to be an electron, then X is a useful fiction for understanding the consequent state of affairs.  CE rests in between these two ideas. C E makes no commitment to the ontic status of the unobservable and can sway the ontic pendulum either way.