Duality and ITR are much less empirical than other aspects of socionics. The ordering of socionics functions are much more theoretically rigorous and therefore falsifiable than MBTI or other similar adaptations of Jung's typology, but besides Olga Tangemann, no one has gotten close to actually empirically testing them. Questionnaires are valid but also have limited validity in applications. ITR would be best tracked through case studies and to a lesser extent controlled experiments, after the participants are typed individually, though double blind experiments with personalities seems extremely difficult to say the least.
Maybe they're "standard procedure" in psychology. Which is why the entire field has less than 40% replication rate...
That they are, but we can't possibly empirically observe all of people's behaviors and thoughts in all situations and all contexts. So we're only studying a minuscule portion of people's behaviors and thoughts.
So it's best to enter the realm of possibilities, and what kind of behaviors and thoughts are possible, not just to study what we have empirically observed so far. THAT, we can test.
The wording was strange in your post before this. If you believe something to be a thing, you should know “what” it is to some extent, even if not in a perfectly complete way empirically. It’s foreign to me at least that people might believe in something they can’t “see” empirically to some degree.